SUICIDE AMONG OLDER ADULTS: AN EXPLORATION OF THE EFFECTS
OF PERSONALITY AND SELF-ESTEEM ON THWARTED BELONGINGNESS,
PERCEIVED BURDENSOMENESS, AND SUICIDAL IDEATION
by
ALLISON EADES
B.A. Concordia University, 2004
M.A. Concordia University, 2009
A thesis submitted to the Graduate Faculty of the
University of Colorado Colorado Springs
in partial fulfillment of the
requirements for the degree of
Master of Arts
Department of Psychology
2016
This thesis for the Master of Arts degree by
Allison Eades
has been approved for the
Department of Psychology
by
Daniel L. Segal, Chair
Fred Coolidge
Leilani Feliciano
07/20/16 Date
ii Eades, Allison. (M.A., Psychology)
Suicide Among Older Adults: An Exploration of the Effects of Personality and Self-
Esteem on Thwarted Belongingness, Perceived Burdensomeness, and Suicidal Ideation
Thesis directed by Professor Daniel L. Segal
Suicide among older adults is a global public health concern. The interpersonal psychological theory of suicide (IPTS) has demonstrated that Thwarted Belongingness
(TB) and Perceived Burdensomeness (PB) are significant predictors of suicidal ideation
(SI). Though much research has been done on SI and factors that contribute to TB and
PB, there is little empirical data exploring how these factors are connected to personality traits and self-esteem. This correlational study explored the relationships and predictive ability of normal and abnormal personality traits (i.e., Five Factor Model - FFM, DSM-5
Personality Disorder traits), and self-esteem on PB and TB in a sample of 102 adults 60 years or older. All five FFM traits were significantly correlated with SI, TB, and PB, and nine of the 10 PD traits correlated significantly with SI, TB, and SI. Linear models using
FFM and PD traits as predictors were significantly predictive of SI, TB, and PB. Results from this study confirm that there is an important relationship between personality and SI and that this relationship may be mediated through TB and PB. This study also confirmed that there is a significant negative relationship between self-esteem, SI, TB, and PB.
Together, results from this study can be used to fill gaps in the literature related to personality, self-esteem, and SI among older adults. This knowledge will have important public health and clinical implications as it can be used to improve screening and treatment for older adults experiencing—or at risk for— SI.
Keywords: suicide, older adults, interpersonal psychological theory of suicide,
thwarted belongingness, perceived burdensomeness, self-esteem
iv
TABLE OF CONTENTS
CHAPTER 1. INTRODUCTION ...... 1
Suicide ...... 2
Suicidal ideation ...... 3
Suicidal intent ...... 3
Suicidal behavior ...... 5
Interpersonal-Psychological Theory of Suicide ...... 5
Thwarted belongingness ...... 7
Perceived burdensomeness ...... 7
Acquired capability ...... 8
Suicidality in Older Adults ...... 9
Rates of suicide ...... 9
Psychological autopsies ...... 10
IPTS and older adults ...... 12
Self-Esteem and Older Adults ...... 15
Personality and Late-life Suicide ...... 16
Personality and the IPTS ...... 18
Summary ...... 21
Present Study ...... 21
2. METHODS ...... 24
Participants and Recruitment ...... 24
Research Design ...... 25
Procedure ...... 25
Measures ...... 27
Geriatric Suicide Ideation Scale ...... 27
Interpersonal Needs Questionnaire ...... 27
Coolidge Axis II Inventory ...... 28
Big Five Mini-Markers ...... 29
Rosenberg Self-Esteem Scale – Bachman Revision ...... 29
Geriatric Depression Scale Short Form ...... 30
Demographic information ...... 31
3. RESULTS ...... 32
Descriptive Statistics ...... 32
Reliability Measures ...... 34
Statistical Analyses ...... 35
Hypothesis 1 ...... 35
Hypothesis 2 ...... 37
Hypothesis 3 ...... 39
Hypothesis 4 ...... 41
Hypothesis 5 ...... 47
Exploratory Analyses ...... 48
Mediation of personality and suicidal ideation ...... 48
Cognition ...... 49
4. DISCUSSION ...... 53
vi
General ...... 53
Personality ...... 55
FFM ...... 55
Personality disorder traits ...... 57
FFM vs. personality disorder traits ...... 60
Self-Esteem ...... 60
General self-esteem ...... 60
Components of self-esteem ...... 61
Neuropsychology and Suicidality ...... 62
Limitations ...... 64
Clinical and Practical Implications ...... 66
Conclusions ...... 66
REFERENCES ...... 68
APPENDIX ...... 76
vii TABLES
TABLE 1. Participant Characteristics ...... 26
2. Psychometric Properties of the Major Study Variables ...... 33
3. Pearson Correlations Between Personality Traits with Suicide Ideation, Perceived Burdensomeness, and Thwarted Belongingness ...... 36
4. Pearson Correlations Between Self-Esteem with Suicidal Ideation, Perceived Burdensomeness, and Thwarted Belongingness ...... 39
5. Linear Regression of Five Factor Model Dimensions on Suicidal Ideation ...... 42
6. Bootstrapped Linear Regression of Five Factor Model Dimensions on Perceived Burdensomeness ...... 42
7. Bootstrapped Linear Regression of Five Factor Model Dimensions on Thwarted Belongingness ...... 43
8. Linear Regression of Five Factor Model Dimensions on Total INQ Score ...... 44
9. Linear Regression of Personality Disorder Traits on Suicidal Ideation ...... 45
10. Bootstrapped Linear Regression of Personality Disorder Traits on Perceived Burdensomeness ...... 46
11. Bootstrapped Linear Regression of Personality Disorder Traits on Thwarted Belongingness ...... 46
12. Linear Regression of Personality Disorder Traits on Total INQ Score ...... 47
13. Pearson Correlations Between CATI Neuropsychological Variables with Suicidal Ideation, Perceived Burdensomeness, and Thwarted Belongingness ...... 50
14. Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Suicidal Ideation ...... 50
15. Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Perceived Burdensomeness ...... 51
16. Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Thwarted Belongingness ...... 52
ix
FIGURES
FIGURE 1. Assumptions of the Interpersonal-Psychological Theory of Suicide. ....6
x CHAPTER 1
INTRODUCTION
With advances in medical technology and improved living conditions, life expectancy has climbed steadily over the last century in North America (Centers for
Disease Control and Prevention, 2013; Statistics Canada, 2010). Concomitantly, the largest demographic group of North Americans—the Baby Boomers—is quickly approaching the age of 65. Over the next twenty years, the proportion of individuals 65 years and older is expected to approach 25% in Canada and 20% in the United States
(Centers for Disease Control and Prevention, 2013; Statistics Canada, 2010). Worldwide, the highest rates of suicide are found among individuals over the age of 65 (World Health
Organization, 2014). An analysis of cohort suicide rates over the past decade in the U.S. revealed that Baby Boomers have had consistently higher rates of suicide compared to other birth cohorts (Phillips, Robin, Nugent, & Idler, 2010). Phillips et al. note that suicide rates are higher among unmarried individuals, and observed an increased in unmarried individuals, which could potentially account for higher rates of suicide among the baby boomers. Other potential explanations include a higher history of substance use among this cohort, and that individuals 50-64 were most affected by the 2008 economic crisis and expected to have difficulty affording retirement. Beyond the cohort differences in suicide rates, Phillips et al. also observed an upswing in suicide rates for all previous cohorts after they reached the age of 60. Consequently, it is believed that suicide rates for 2 seniors will continue to grow as the Baby Boomers reach older age. Even though researchers have been predicting increases in suicide rates among older adults for years, the risk of suicide in this group has been downplayed, and interventions in the U.S. mostly target youth (Fiske, 2014). Recognizing these facts, suicide among older adults should be considered as an important public health problem. In order to effectively nurture the next generations of seniors, it is critical that we understand the intricacies of suicide among older adults. The purpose of this study was to explore the relationship— and predictive ability—of personality and other psychological factors on thwarted belongingness, perceived burdensomeness, and ultimately suicidal ideation.
Suicide
Generally, suicide refers to an intentional self-inflicted act that results in one's own death. Though completed suicide is defined by an end result of death, the concept of suicide encapsulates thoughts (ideation), objectives (intent), and behaviors (action) related to a self-inflicted death and/or self-harm (Beck, Schuyler, & Herman, 1974).
Despite the importance of studying suicide, there are some major methodological challenges in doing so. First, those who have completed suicide are not available to complete questionnaires or interviews. Though psychological autopsies may be performed to develop an understanding of the medical, social, and psychological factors that may have contributed to one's suicide (INSERM Collective Expertise Centre, 2000), this information is secondhand, retrospective, and can be limited regarding the scope and accuracy of data that can be uncovered. Not having direct access to the suicidal individual places limits on the research methods that can be conducted (Van Orden et al.,
2010).
3
Second, from an ethical and safety perspective it is difficult to conduct research with individuals who are actively suicidal. Actively suicidal individuals are usually in crisis, which requires immediate intervention; consequently it unethical to delay crisis support for assessment purposes. Recognizing these methodological challenges, suicide research is most often conducted with individuals from a retrospective perspective (i.e., reflecting on how they were feeling at the time of their suicidal behaviors). Researchers may also choose to conceptualize suicidal ideation and behavior as dimensional concepts that exist to a lesser degree within the general population. Despite the methodological challenges, suicide is an important research topic and a better understanding of it could potentially mitigate a public health crisis. In order to study suicide in a methodologically sound manner, we must first outline and define the components of suicide.
Suicidal ideation. Suicidal ideation (SI) is comprised of thoughts and desires concerning one's own death. It should be noted that these thoughts are separate from overt suicidal behaviors, in that one may or may not take action to make these thoughts a reality (Beck, Kovacs, & Weissman, 1979). Suicidal ideation may include private thoughts, or "threats" vocalized to others. Furthermore, suicidal ideation may be either passive (i.e., belief that life is not worth living, and/or wishing for death) or active (i.e., thinking about ways to actively end one's own life). Passive and active SI are considered to be part of a spectrum of intensity; thus SI is not a static concept, and may fluctuate in frequency and intensity across one's life course (Szanto et al., 1996; Van Orden et al.,
2014).
Suicidal intent. Suicidal intent represents the seriousness and the intensity of one’s desire to end his/her own life. The two primary components of suicidal intent are
4 one’s desire for death and one's estimate of the effectiveness of his/her plan for suicide
(Beck et al., 1974). The desire for death is, in fact, a struggle between one's desire to no longer be living and one's innate instinct for self-preservation (Beck et al., 1974). When the self-preservation instinct is low and desire for death is high, suicidal intent will also be high and the resulting suicidal attempts may be more serious. When the self- preservation instinct is equal to the desire for death, individuals are more likely to feel ambivalent about death and their actions are likely to reflect this ambivalence.
The other component of suicidal intent is an estimate of effectiveness. This concept reflects one’s ability to accurately estimate the likelihood that one’s plan for suicide will result in death (Beck et al., 1974). Several factors appear to influence one's probability estimate. Past suicide attempts allow the individual to reflect on the reasons that the plan was unsuccessful and factors that could be modified to increase the likelihood of death. Similarly, exposure to the suicide of a relative or a close friend may also provide similar knowledge about completing suicide. It should be noted however, that an accurate estimate of the likelihood of death is separate from estimating lethality
(Beck et al., 1974). Though using a gun to inflict a wound to one’s head is highly lethal, its likelihood of death could be lower if the plan is to hold a gun to one’s head in front of family members while saying goodbye compared to doing so in isolation. In this example, individuals may inaccurately assess the effectiveness of their plan if they assume that they will be able to pull the trigger without being talked-out of it by their loved ones, or that they would not second guess themselves about making their family witness such an event.
5
Suicidal behavior. Another challenge in the study of suicide is reaching consensus on how to define suicidal behavior. Generally, suicidal behaviors represent self-initiated actions that are associated with the intent to kill or harm one's self.
According to Van Orden et al. (2010, p. 2) "these behaviors can vary in terms of the presence or absence of intent to die and presence or absence of physical injury sustained."
When individuals engage in behaviors with no intent to die, but that lead to physical injury, it is referred to self-harm. In order to hone in on individuals with a true wish for death, recent research has defined suicide attempts by the "following qualities: 1) self- initiated, potentially injurious behavior; 2) presence of intent to die; [and] 3) non-fatal outcome" (Van Orden et al., p. 2). Though it is essential to understand the different aspects of suicide, there are many factors that contribute to different levels of ideation, intent, and behavior.
Interpersonal-Psychological Theory of Suicide
A relatively new and popular theory of suicide is the Interpersonal-Psychological Theory of Suicide (Joiner, 2005). Joiner's model has significant clinical implications because it was one of the first models with the ability to predict Suicidal Ideation (SI) and the transition from SI to actual suicide attempts (O’Connor & Nock, 2014). To inform the theory, Joiner and colleagues examined risk factors for suicide that hold high levels of empirical support (i.e., mental disorders, past suicide attempts, social isolation, family conflict, unemployment, and physical illness; Van Orden et al., 2010). Joiner et al. then developed a theory that could account for how diverse risk factors are related to suicidal behavior. The fundamental premise of the IPTS is that suicidal individuals have a desire to die and that they will take action to end their life if they can overcome their innate fear
6 of pain and death (see Figure 1). The constructs of the IPTS that relate to an individual’s wish for death are thwarted belongingness (TB) and perceived burdensomeness (PB); whereas acquired capability (AC) relates to behaviors meant to help achieve death.
Together, the IPTS suggests that when individuals feel unwanted social isolation and perceive themselves as a burden to others, their desire for death is higher and they are more likely to complete suicide if they have also acquired the capability to overcome their innate fear of pain and death (see Figure 1). Joiner and colleagues have completed extensive research to identify and test variables that contribute to TB, PB, and AC.
Though the theory was initially developed and tested with a sample of young adults (i.e.,
18-26 year olds), the constructs have subsequently been tested and validated with older adults (Cukrowicz, Cheavens, Van Orden, Ragain, & Cook, 2011).
Figure 1. Assumptions of the Interpersonal-Psychological Theory of Suicide. Adapted from “The Interpersonal Theory of Suicide,” by K. A. Van Orden, T. K. Witte, K. C. Cukrowicz, S. R. Braithwaite, E. A. Selby, & T. E. Joiner, 2010, Psychological Review, 117, p. 42.
7
Thwarted belongingness. According to the IPTS, thwarted belonging results from one's unmet need to belong. Numerous studies demonstrate that "social isolation is arguably the strongest and most reliable predictor of suicidal ideation, attempts and lethal suicidal behavior among samples varying in age, nationality, and clinical severity" (Van
Orden et al., 2010, p. 5). Thus TB is an important construct to consider when studying suicide.
According to Van Orden et al. (2010), TB is comprised of two latent variables, loneliness (i.e., too few social connections) and an absence of reciprocal care (i.e., feeling that you care for others and that you are cared for by others). These variables may impact belongingness in different ways, yet combined, they encapsulate one's global experience of belongingness by addressing one's desire for social connection and how connected one feels to others. According to Van Orden et al. (2010), observable indicators of loneliness include: self-reported loneliness, pulling-together effects, marriage and number of children, seasonal variations, living alone, and few social relationship, or little familial interaction. Observable indicators for absence of reciprocal care are: social withdrawal, low openness to experience, residing in a single jail cell, domestic violence, childhood abuse, and family conflict. TB is a dynamic dimensional concept, which means that feelings of belongingness are not static, and they may exist to varying degrees and different moments in time.
Perceived burdensomeness. When developing the IPTS, researchers attempted to create constructs that explained and linked various risk factors and predictors of suicide together. TB as a construct was developed based on established research about social isolation and its relationship to suicide, however PB was developed from trying to
8 identify the common factor between family conflict, unemployment, and physical illness
(Van Orden et al., 2010). According to the IPTS the common thread between the aforementioned risk factors and suicide, is feeling like a burden to significant others.
As reported by Van Orden and colleagues (2010), perceived burdensomeness is comprised of two latent constructs, liability (i.e., my death is worth more to others than my life), and self-hatred. According to Van Orden et al., observable indicators of the liability construct include distress from: homelessness, incarceration, unemployment, and physical illness, in addition to thoughts of expendability and that one is a burden on their family. Indicators of self-hatred include: low self-esteem, self-blame/shame, and agitation. "As with thwarted belongingness, perceived burdensomeness is presumed to be a dynamic cognitive affect state, as well as a dimensional phenomenon" (Van Orden et al., 2010, p. 13). Thus, the intensity of PB that individuals experience can vary across time and within different relationships. Van Orden et al., believe that PB is at its most critical level when an individual perceives himself or herself as a burden to multiple significant others and when the experience of burden is accompanied by self-hatred.
Acquired capability. Acquired capability (AC) is the idea that one's fear of death can be reduced by repeated exposure to events that desensitize one's self to pain (e.g., tattoos/piercings, cutting, torture) and to fear (e.g., combat experience, repeated exposure to death, previous suicide attempts). These exposures results in a diminished fear of death and an increased capacity to undertake lethal suicidal behaviors—something that is both against our evolutionary drive for self-preservation and inherently difficult due its frightening and painful nature (Beck et al., 1979; Joiner, 2005; Van Orden et al., 2010).
Though AC is conceptualized as something that is developed through repetitive exposure,
9 it is important to note that individual differences in factors such as personality, impulsivity, pain tolerance, and fearlessness may exist, which could increase one's ability to develop AC (Van Orden et al., 2010). This suggests that some individuals may be predisposed to the ability to complete suicide, whereas others may required several attempts before their suicidal behavior results in death.
Suicidality in Older Adults
Rates of suicide. Suicide among older adults is an important, but often overlooked public health problem. The Centers for Disease Control and Prevention in the
United States (US) estimates that each suicide costs approximately 1 million dollars in medical and work lost costs; this translates into approximately 35 billion dollars in the
US annually (Centers for Disease Control and Prevention, 2014b). Though many older adults are retired, their suicides may still have costs related to work loss for those they have left behind (e.g., children). In 2012, suicide accounted for 1.4% of all deaths worldwide, and was the 15th leading cause of death (World Health Organization, 2014).
In the US, as of 2013, suicide was the 10th leading cause of death (Centers for Disease
Control and Prevention, 2014a). Globally, individuals 70 years of age and older have the highest rates of suicide among most nations (World Health Organization, 2014). In 2012, the estimated suicide rate in the US for individuals over the age of 70 was 16.5 per
100,000, which represents the second highest rate by age group, the highest rate of 20.1 per 100,000 belonging to the adjacent age group (50-69 years of age; World Health
Organization, 2014). This data supports findings that the Baby Boomers have high rates of suicide (Phillips et al., 2010) and that suicide rates for older adults in the US are likely to increase in the coming years. Though the aforementioned suicide rate estimates do not
10 specify the reasons for suicide, contrary to what one may assume, research explored later in this paper suggests that suicide among older adults is not primarily related to physical ailments and/or declining health.
In order to truly understand suicide among older adults and how it may or may not relate to the IPTS, we must review research that outlines various risk factors and commonalities between older suicide completers.
Psychological autopsies. Psychological autopsy is an investigative procedure used after an individual's death to estimate his/her pre-mortem psychological state and level of mental wellbeing. It generally involves interviews with family members and access to medical records and/or personal documents in order to explore how the deceased person was feeling, what they were thinking, and what they had been doing before their death. The goals of this type of research are to try to understand more clearly why the individuals completed suicide, and to discover, perhaps, what could have been done to prevent the deaths.
According to a review of psychological autoposy studies conducted by Cavanagh,
Carson, Sharpe, and Lawrie, (2003), the strongest common variable among individuals who complete suicide was the presence of a diagnosed mental disorder (observed in 90% of the cases reviewed, with depression being the most common diagnosis). However, research conducted by De Leo, Draper, Snowdon, and Kõlves (2013) suggests that even though individuals who die by suicide are more likely to have a mental disorder
(compared to those who die by natural causes), older adults who die by suicide had a significantly lower prevalence of mental health disorders compared to suicides completed by middle aged individuals (61.6% compared to 80.1%).
11
Though it might be tempting to assume that late-life suicide is the response to declining physical health, the relationship may not be so direct. Unsurprisingly, older adults in De Leo’s study who died by suicide had significantly higher physical illness ratings compared to their middle-aged suicide counterparts. However, the late-life suicide group had lower illness rating scores compared to the late-life control group.
Furthermore, the late-life sudden-death group demonstrated significantly greater impairment in physical self-maintenance (i.e., inability to physically care for oneself) compared to the older suicide completers. Together, these findings suggest that the older adult suicide group had a similar levels of physical illness compared to the non-suicide group but that they had less impairment in their ability to care for themselves; despite this, they still chose to complete suicide. De Leo’s data may suggest that physical health and/or impaired self-care were not a major reason for suicide among older adults. Along these lines, Turvey et al. (2002) state that even though the prevalence of significant illnesses increases with age, few individuals choose suicide. Notably, Turvey et al. found no significant differences “between control subjects and suicide victims in the number and type of chronic medical illnesses and cognitive or physical impairment” (p. 404).
Other research acknowledges that physical illness may contribute to suicide especially when the illness results in chronic pain (e.g., osteoarthritis, cancer - Harwood, Hawton,
Hope, Harriss, & Jacoby, 2006). However, Harwood et al. report that this relationship may be moderated through depression. This relationship is important to consider given that major depression is recognized as the most salient risk factor for suicide among older adults (Turvey et al., 2002).
12
Additional results from De Leo et al.’s (2013) study of psychological autopsies suggest that dementia and other cognitive disorders do not appear to be an important factor in suicide in older adults. They found that only 1.4% of the suicide group compared to 9% in the sudden death group had a diagnosis of dementia and other cognitive impairment. Beyond the biological and psychological, De Leo et al. discovered that older adults who died by suicide had significantly less social support from family and friends compared to controls and that living alone was a significant predictor of suicide in older adults. These findings suggest that suicide among older adults must be conceptualized from a biopsychosocial perspective as opposed to solely from a biological or psychological perspective.
IPTS and older adults. As many individuals age, they become more reliant on others to meet their instrumental activities of daily living (e.g. banking groceries, transportation) and activities of daily living (e.g. cooking, dressing, bathing). At the same time, many older adults are socially vulnerable because of their shrinking social circle, which may result from the death of friends/spouses, reduced mobility due to functional decline, or the need to relocate to a new city to be closer to children who can provide additional support with instrumental activities of daily living. With TB and PB as central constructs, the IPTS lends itself well to the study of suicide in older adult populations.
The Interpersonal Needs Questionnaire (INQ) is a popular self-report measure designed to assess TB and PB. A growing body of research has demonstrated strong validity for the INQ in older adults (Marty, Segal, Coolidge, & Klebe, 2012) and has demonstrated that TB and PB can accurately predict SI among older adults. Specifically,
Van Orden, Cukrowicz, Witte, and Joiner (2012) used a test of equal form to measure the
13 latent structures of the INQ (developed with a sample of undergraduate students) against a sample of older adults; their results confirmed the equivalence of latent structures between samples. Furthermore, Cukrowicz, Cheavens, Van Orden, Ragain, and Cook
(2011) found that PB accounts for a significant amount of variance in SI among older adults, even when controlling for depressive symptoms, hopelessness, and functional impairment. Furthermore, a case controlled sample of suicide completers (via psychological autopsy) with living controls 50 years of age or older, also confirmed that risk for burdensomeness, and painful and provocative experiences are significant predictors of suicide among older adults even when major depression is controlled for
(Van Orden, Smith, Chen, & Conwell, 2016). Together, these studies confirm the appropriateness of the IPTS as a theory for conceptualizing suicide among older adults and the INQ as a way to measure TB and PB in this population.
Lack of belongingness and social connection (latent variables of TB) are recognized risk factors for suicidal ideation, however there is little research examining the direct relationship between perceived burdensomeness and suicidal ideation. An unpublished meta-analysis examining the effect size of PB on SI in older adults, uncovered five quantitative studies that measured the effect of PB on SI. The standardized effect sizes in these studies ranged from r = .59 to r = .80, and the meta- analysis revealed a pooled effect size of .72, (SE = 0.04) 95% CI [0.63, 0.80] (Eades,
2014). These data suggest that PB has a large effect on SI, and that as PB increases, so does SI.
Beyond quantitative research, there have been some qualitative studies that have analyzed the content of suicide notes from completed suicides for evidence of PB.
14
Though not exclusively focused on suicide by older adults, Foster (2003) reported that notes written by older adults (65 and older) were significantly more likely to contain themes of burdensomeness than those of younger and middle aged adults (40% vs. 3%, p
= .03). Contrary to these findings, a review of suicide notes studies which included a total of 664 suicide notes, detected themes of TB in 42.5% of the notes but only detected themes of PB in 15.5% of the notes (Lester & Gunn, 2012). Lester and Gunn also observed that men were less likely to write about burdensomeness than women, and that older adults wrote less often about TB than their younger counterparts. Though age data was not available for the all the suicide note samples in Lester and Gunn's study, the range of means reported was between 22.4 (SD = 3.4) to 44.8 (SD = 6.8) years of age. In recognition of their findings, Lester and Gunn contend that their study refutes the IPTS' claim that elements of TB and PB must be present for suicide to occur. A recent qualitative study queried a sample Swedish suicide survivors 70 years of age or older upon their admission to the emergency department about the reason for their suicide attempt. The study reported that the most common attributions behind the non-lethal suicide attempts were a desire to escape life, somatic problems and pain, and psychological difficulties (Van Orden et al., 2015). In fact, the authors of this study acknowledge that TB and PB were not the most commonly reported reasons for the suicide attempt and suggest that the explanatory power of the IPTS might be limited in certain cultural or clinical profiles.
Christen, Batterham, Mackinnon, and Soubelet (2013) also found data that refutes part of the IPTS. Using data from three waves of a longitudinal study, Christen et al. reported that among their sample of 40 year olds, TB did not predict SI. These results
15 support Lester and Gunn’s argument against the central premise of the ITPS. However, when Christensen et al. analyzed data from their entire sample, which consisted of three different age groups (20s, 40s, and 60s) they were able to detect a significant interaction between PB and TB in relation to SI. Using Christensen et al.’s (2013) data plus a fourth- wave of data, Christensen, Batterham, Mackinnon, Donker, and Soubelet (2014) replicated these findings. Thus, Christensen et al. (2014) acknowledge the predictive ability of the IPTS for SI and completed suicide, while also noting instances where the
ITPS' theory-based predictions do not hold true.
Self-Esteem and Older Adults
There is little modern research discussing the relationship between self-esteem with suicide, and there is even less research looking at the self-esteem of older adults or the relationship between suicide and self-esteem in older adults. Considering that Van
Orden et al. (2010) recognize self-esteem as an indicator of self-hatred (a variable of PB) the link between self-esteem and PB merits exploration.
Kjølseth, Ekeberg, and Steihaug (2009) found that in a sample of older
Norwegian suicide attempters, self-esteem was associated with a sense of achievement; the perceived loss of this ability resulted in the self-perception that they held no personal value. Harrison et al. (2010) noted that older individuals with severe SI and/or suicidal attempts were more likely to report "low perceived social support, as manifested in the lack of a sense of belonging, tangible support, and self-esteem" (p. 6).
The most in-depth study of SI and self-esteem among older adults demonstrated a significant negative relationship between PB and self-esteem (Marty, 2011). Despite the fact that the IPTS does not identify self-esteem as a variable of TB, Marty observed a
16 stronger correlation between self-esteem and TB than with PB; she noted that these findings could have been caused by a lack of sensitivity in the 1-item self-esteem measure used. Marty also proposed that these findings could mean that self-esteem is not in fact an indicator of self-hatred among older adults. Recognizing that belongingness is a central component of self-esteem (Rosenberg, 1965), it could also be conceivable that TB is highly correlated with self-esteem but that self-esteem does not have a causal relationship with TB. Furthermore, it is possible that self-esteem among older adults is more strongly related to the variables of TB (i.e., loneliness and absence of reciprocal care) than to the variables of PB (liability and self-hatred). Currently there is a paucity of research on self-esteem among older adults, thus it is unclear how older adults conceptualize their self-esteem. Consequently, a more detailed investigation is required to better understand what self-esteem among older adults is comprised of, and how it relates to TB, PB, SI.
Personality and Late-Life Suicide
The relationship between personality factors and suicide has been well established within a rich research tradition. Several studies, using Costa and McCrea's (1992) Five
Factor Model (FFM) of personality, have demonstrated a link between suicide and
Neuroticism, Extraversion, Openness to Experience, Agreeableness, and
Conscientiousness across the life span. Specifically, high levels of Neuroticism are consistently associated with high levels of SI among younger and older adults (Chioqueta
& Stiles, 2005; Cramer et al., 2012; DeShong, Tucker, O’Keefe, Mullins-Sweatt, &
Wingate, 2015; Heisel & Flett, 2006; Iliceto, Fino, Sabatello, & Candilera, 2014; Segal,
Marty, Meyer, & Coolidge, 2012; Useda, Duberstein, Conner, & Conwell, 2004), low
17
Extraversion is also commonly associated with higher levels of SI among younger and older adults (Chioqueta & Stiles, 2005; Cramer et al., 2012; DeShong et al., 2015; Iliceto et al., 2014; Segal et al., 2012), whereas Openness to Experience, Agreeableness, and
Conscientiousness show a small, or no relationship, with SI among older adults (Heisel &
Flett, 2006; Segal et al., 2012; Useda et al., 2004). Among older adults, the most prominent FFM research has demonstrated that individuals at risk for suicide often have low Openness to Experience, high levels of Neuroticism, and/or low levels of extraversion (Conwell, Duberstein, & Caine, 2002; Duberstein et al., 2000; Duberstein,
Conwell, & Caine, 1994; Useda et al., 2004).
Beyond the normative FFM personality dimensions, diagnosable personality disorders (PD) are important considerations for the mental health of older adults (Jahn,
Poindexter, & Cukrowicz, 2015; Segal et al., 2012). According to the DSM-5 (American
Psychiatric Association, 2013), PDs are pervasive and inflexible patterns of statistically infrequent inner experiences and behaviors that lead to distress and impairment in daily activities and social relationships. Though PDs can be impactful at any stage of life, PDs are especially significant for older adults as they are faced with a new phase of life that— for many—can be characterized by changes in physical health and diminished social networks. Without effective coping strategies, older adults with PDs are at risk for developing comorbid depression and anxiety and are therefore at greater risk for suicide
(Harwood, Hawton, Hope, & Jacoby, 2001; Heisel, Links, Conn, van Reekum, & Flett,
2007; Segal, Coolidge, & Rosowsky, 2006). According to Reynolds, Pietrzak, El-
Gabalawy, Mackenzie, and Sareen (2015), 14% of older Americans have at least one diagnosable PD. These data are aligned with rates previously reported by Segal et al.
18
(11% - 2006) and Abrams and Horowitz (10% - 1996). There is also evidence that over
40% of suicide completers over the age of 60 have a diagnosable PD or PD traits
(Harwood et al., 2001; Segal et al., 2012). Furthermore, PDs account for over 50% of the variance in SI among older adults (Segal, Gottschling, Marty, Meyer, & Coolidge, 2015).
Though the prevalence of PDs is generally lower among older adults compared to younger adults (10-14% vs. 18-21% respectively, Segal et al., 2006) PDs in late-life are of particular concern because older adults with PDs have often gone through life with poor coping skills, which can result in repeated failures and diminished self-esteem. In older adulthood these individuals are often faced with stressors such as physical and cognitive declines, limited social networks, reduced independence, and restricted financial freedom, which can be very difficult to cope with especially if one has poor self-esteem (Segal et al., 2006). Consequently, they are more likely to experience comorbid mental health diagnoses such as depression which is strongly related to late-life suicide. Thus the link between PDs, older age, and suicide is an important area of exploration.
Personality and the IPTS. Though several studies have identified the personality-related risk factors of suicide for older adults, there is less research on the relationship between personality and the core constructs of the IPTS. One study explored potential predictive factors of TB, PB, and AC using Eysenck's personality traits of
Neuroticism, Extraversion, and Psychoticism (Christensen et al., 2014). These researchers found that low Neuroticism was predictive of higher levels of AC and higher
Psychoticism was predictive of TB and AC. No connections between Extraversion and the IPTS constructs were reported. Furthermore, the three personality traits measured did
19 not have any significant predictive ability for PB (Christensen et al., 2014). Meanwhile,
DeShong et al. (2015) discovered a strong positive relationship between Neuroticism and
TB and a negative relationship between TB and Extraversion and Agreeableness
(DeShong et al., 2015). Additionally, there is evidence of a positive relationship between
PB and Neuroticism (Cramer et al., 2012; DeShong et al., 2015) and a negative relationship between PB and Extraversion (DeShong et al., 2015).
Though Silva, Ribeiro, and Joiner (2015) did not directly measure personality traits, their research explored the relationship between different DSM-IV Axis I and Axis
II disorders and the constructs of the IPTS. They found that TB was significantly predicted by depressive disorders, bipolar disorders, social phobias, polysubstance dependence, and borderline personality disorder (Silva et al., 2015). PB was significantly predicted by depressive disorders, bipolar disorders, polysubstance dependence, schizophrenia and other psychotic disorders, somatoform disorders, schizotypal personality disorder, borderline personality disorder, and personality disorders NOS.
Acquired capability was significantly predicted by PTSD, inhalant abuse/dependence, schizophrenia or other psychotic disorders, and schizotypal personality disorder.
Unsurprisingly, avoidant personality disorder was a significant negative predictor of AC
(Silva et al., 2015); this is likely because individuals with avoidant personality disorder are often reluctant take risks or engage in new activities that could result in failure or embarrassment (e.g., an incomplete suicide attempt).
In the one published study of personality and the IPTS focusing on older adults,
Jahn et al. (2015) used the Structured Clinical Interview for DSM-IV-TR Axis II
Personality Disorders, Patient Questionnaire (SCID-II-PQ) to obtain a total number of
20 personality traits. Their study aimed to explore the relationship between personality traits and suicidal ideation as mediated by depressive symptoms, hopelessness, PB and TB.
Total PD traits were found to have a significant relationship with SI, accounting for 43% of the variance. Interestingly, the relationship was no longer significant after the mediating variable were added. Despite this, depressive symptoms, as well as TB and PB were significant mediators for the relationship between personality traits and SI (Jahn et al., 2015). This study demonstrates an important relationship between personality and TB and PB, however it does not elucidate which personality traits have what relationship with PB and TB.
Eades, Segal, Marty, and Coolidge (2015) examined the relationship between specific PD traits, TB, and PB among community-dwelling older adults, and found significant correlations. Specifically, TB was positively correlated with self-defeating, avoidant, depressive, paranoid, Schizoid, Passive-Aggressive, Antisocial, Dependent,
Sadistic, Obsessive-Compulsive, Schizotypal, and Histrionic PD traits. Of those, moderate effects were found for Self-Defeating, Avoidant, Depressive, Paranoid,
Schizoid, Borderline, Passive-Aggressive PD traits. PB was significantly correlated with the same PDs as TB, with the exception of Histrionic PD. Moderate effects were detected between PB and self-Defeating, Depressive, Schizoid, Avoidant, Paranoid, Borderline, and Passive-Aggressive PD traits. Though similar in nature to the Eades, Segal, Marty, and Coolidge (2015) study, our current study aimed to explore these relationships in greater detail. Specifically, the aforementioned study assessed personality disorder traits using the 70-item Short Coolidge Axis II Inventory (SCATI), whereas our current study opted to use the full 250-item Coolidge Axis II Inventory (CATI). In doing so there are
21 more items measuring each PD, which should increase reliability and validity.
Furthermore, the full CATI has additional subscale related to cognitive functioning, which allowed us to complete some exploratory analyses with regards to the relationship between suicidal ideation and cognitive impairment among older adults.
Summary
Though there is evidence of TB and PB as predictors of suicide for older adults, there is a paucity of research examining the relationship between personality and suicide within a theoretical construct (e.g., IPTS). Specifically, little is known about how personality affects PB and TB among older adults, nor is there much evidence about the psychological factors (e.g., self-esteem, loss of social/personal worth) that predict PB in older adults. Furthermore, there is potentially contradictory evidence about the salience of PB in older suicide completers. Thus, research is required to elucidate variables that correlate with, and predict, PB in older adults while simultaneous replicating previous findings with regards to TB, SI, and older adults.
Present Study
The purpose of this study was to identify normal and abnormal personality traits and other relevant psychological factors that correlate with—and that predict—PB and TB in older adults. Furthermore, in taking into consideration the latent constructs of TB and PB, this study explored factors such as personality traits, personal/social worth, and self- esteem to identify how they might relate to and predict TB, PB and SI.
The hypotheses were as follows:
1. FFM personality traits would be significantly correlated with TB, PB, and SI.
22
a. TB was expected to correlate positively with Neuroticism, but would
correlate negatively with Openness to Experience, Conscientiousness,
Extraversion and Agreeableness.
b. PB was expected to correlate positively with Neuroticism and
correlate negatively with Openness to Experience, Conscientiousness,
Extraversion, and Agreeableness.
2. PD traits would be significantly correlated with TB, PB, and SI.
a. TB was expected to correlate positively with Self-Defeating,
Avoidant, Depressive, Paranoid, Schizoid, Passive-Aggressive,
Antisocial, Dependent, Sadistic, Obsessive-Compulsive, Schizotypal,
and Histrionic PD traits.
b. PB was expected to correlate positively with Self-Defeating,
Avoidant, Depressive, Paranoid, Schizoid, Passive-Aggressive,
Antisocial, Dependent, Sadistic, Obsessive-Compulsive, and
Schizotypal PD traits.
3. PB and TB would have significant positive correlations with the loss of
personal and social self-worth subscale of the GSIS (GSIS-PSW), and PB
and TB would have significant negative correlations with self-esteem. It was
expected that TB would correlate more strongly with the positive self-regard
dimension of self-esteem and that PB would correlate more strongly with the
usefulness/competence dimension of self-esteem.
4. PD traits and FFM personality traits would significantly predict TB, PB, and
SI.
23
a. The overall model for PDs traits would account for more variance in
TB, PB, and SI compared to the overall model of FFM traits.
5. GSIS-PSW and self-esteem would be predictive of TB, PB, and SI after
controlling for depression.
6. We also conducted exploratory analyses as described next, but without
hypotheses.
a. We explored the extent to which self-esteem, TB, and PB mediate the
relationship between PDs traits/FFM traits and SI.
b. We explored the relationship between neurocognitive dysfunctions,
TB, PB, and SI.
CHAPTER 2
METHOD
Participants and Recruitment
A convenience sample of adults 60 years of age or older was recruited through the
University of Colorado Colorado Springs research registry database. All individuals on the registry list over the age of 60 who included an email address were sent an email invitation, whereas a selection of individuals with telephone numbers only were called.
Participants were given the choice of completing the study online, in person at the
UCCS-Aging Center, or having the research packet sent in the mail to them (with return postage). In total, 97 individuals chose to complete the questionnaires online, three individuals chose to complete the study in person, whereas 15 individuals requested and returned a paper copy of the research packet. To compensate individuals for their time, participants were offered the opportunity to enter in a raffle for one of two $25 VISA gift cards. Seventy-seven individuals entered their name in the raffle and two winners were selected by using a random number generator.
In total 110 individuals completed the research packet; data for two individuals were removed from the analysis because they were under the age of 60, and data from 6 other individuals were also removed because the participants failed to complete at least two thirds of the research packet. Of the 102 cases retained, approximately five cases had some missing values. A Little’s Missing Completely at Random (MCAR) test confirmed 25 that the data was missing at random. Thus, Expectation Maximization was conducted in
SPSS as a way to impute the missing values. The final sample consisted of 102 participants; their age ranged from 61-93 years (M = 73.26 years, SD = 5.69 years). The gender distribution was biased toward women (72.5%, n = 74) over men (26.5%, n = 28) but was proportional to the ratio of women to men on the research registry (71% vs.
29%). The ethnic composition of the sample reflected non-Hispanic or Latino individuals
(95.1%, n = 97) and individuals from a Hispanic or Latino heritage (4.9%, = 5). Racially the sample identified predominantly as White (94.1%, n = 96), with 1.9% (n = 2) of the sample identifying themselves as being of mixed White and American Indian race.
Nearly 4% of the sample classified their race as “other” or chose not to disclose this information (see Table 1).
Research Design
The proposed study was correlational in nature; as such it aimed to explore the relationship between PB and TB with a variety of naturally occurring psychological and demographic variables. Specifically, this study used simple correlations, and a series of multiple linear regressions to explore the relationship between PB, TB, self-esteem, and
SI. In doing so, this study aimed to advance the field of study relating to suicide risk and resilience in older adults.
Procedure
The study received ethics approval from UCCS' Institutional Review Board
(protocol 16-110), ensuring that it meet all ethical requirements related to the study of human participants. Individuals with email address were sent an email with a description
26
Table 1
Participant Characteristics (N = 102) M SD Range Age 73.3 5.7 61 – 93 Education 16.36 2.5 12 – 20+
n % Gender Female 74 72.5 Male 27 26.5 Unknown 1 1.0
Racial/Ethnic Not Hispanic or Latino 97 95.1 Hispanic/Latino 5 4.9 White 96 94.1 Hispanic 1 1.0 Multiracial (American Indian) 2 1.9 Do not wish to answer/Other 3 2.9
Relationship Married 58 56.9 Widowed 18 17.6 Divorced 16 15.7 Single/Never Married 5 4.9 Domestic partnership 5 4.9
of the study and a link to the survey. The email also provided individuals with instructions about how to proceed if the preferred to complete the study in person or to receive a paper copy of the research packet. Three weeks after the first email was sent, a reminder email was sent to individuals. To increase recruitment numbers and to mitigate the risk of a sample bias toward individuals with internet access, individuals from the research registry with phone numbers only were selected at random and invited to participate in the research. In total 23 individuals were called, voicemails describing the study were left for individuals who did not answer the phone; a total of 12 individuals agreed to take part of the study. Individuals recruited over the phone were offered the 27 same three options for completing the research packet. The research packet comprised six questionnaires, in addition to basic demographic information (e.g., age, living arrangement, marital status). Participants only had to complete the package once.
Participants were told that the expected time to complete the research packet was 60-90 minutes, however the average time to complete the questionnaire online was 47 minutes.
Measures
Geriatric Suicide Ideation Scale (GSIS). The GSIS is a 31-item self-report instrument with a 6-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree) (Heisel & Flett, 2006). Examples of items include: I have recently been thinking a great deal about specific ways of killing myself, I often wish that I would pass away in my sleep; Life is extremely valuable to me. The GSIS has four subscales that can be used as stand-alone measures: Suicide Ideation, Death Ideation, Perceived Meaning in Life, and
Loss of Personal and Social Worth. Higher scores on the GSIS and its subscales represent greater suicidality. This instrument has demonstrated strong reliability and internal consistency with a sample of older adults (Heisel & Flett, 2006; Marty, 2011; Segal et al.,
2012). The total sum score and the sum score for the Loss of Personal and Social Worth subscale (GSIS-LPSW) were analyzed for this study.
Interpersonal Needs Questionnaire – 15-item. The INQ is a popular self-report measure (Van Orden et al., 2012) with demonstrated validity across a variety of samples including older adults (Marty, 2011; Segal et al., 2012; Van Orden, 2009) and is designed to assess TB and PB as defined by the IPTS (Van Orden et al., 2012). It is a 15-item self- report measure that prompts respondents to reflect upon how true the items have been for them recently. The answers are based on a 7-point scale ranging from 1 (not at all true 28 for me) to 7 (very true for me). Examples of items include: These days the people in my life would be better off without me, These days I think that I am a burden on society. Nine items are used to measure thwarted belongingness while the remaining six items measure perceived burdensomeness. Higher scores on the INQ’s subscales represent greater levels of TB and PB. Sum scores for the TB and PB subscales of the INQ, as well as the total
INQ score, were analyzed for this study.
Coolidge Axis II Inventory (CATI). The CATI is a dimensional measure of PD traits that has been validated in both clinical and non-clinical populations of older adults
(Coolidge & Merwin, 1992; Segal et al., 2006, 2012). It is a 250-item self-report measure of the 10 PDs listed in the DSM-IV-TR/DSM-5. Beyond the 10 standard PDs, the CATI measures Depressive, Passive-Aggressive, Self-Defeating, and Sadistic PD traits from earlier DSM versions. The CATI also assesses for six clinical syndromes (e.g., major depressive disorder, generalized anxiety disorder, post-traumatic stress disorder, social phobia, and attention-deficit disorder), and neuropsychological deficits (Coolidge &
Merwin, 1992). Respondents answer each item on a 4-point Likert-type scale ranging from Strongly False, More False than True, More True Than False, to Strongly True.
The raw scores from PD scales and clinical syndromes scales are converted to T-scores; a score of 50 represents the norm with a standard deviation of 10 points. Scores that are two standard deviations above the mean are considered to be clinically significant. For the purpose of this study, the CATI was used to assess the dimensional scores of all PD features and traits; T-scores for the different PD and neurocognitive dysfunction scales were analyzed for this study.
29
The CATI was also used to determine a participant’s validity in responding. The
CATI's three random responding items, total raw scores, and neurocognitive deficit scale were used to flag invalid responses and/or severe cognitive impairment. A score greater than or equal to three on the random response items suggests that the respondent is not being truthful in their reporting; whereas a total raw score on the CATI less than or equal to 130 is considered to be an invalid score because of symptom denial. Furthermore, more than 15 omissions on the neurocognitive dysfunction scale also indicates invalid responding. No participants were excluded based on these three validity checks.
Big Five Mini-Markers (BFMM). Saucier’s (1994) Big Five Mini-Marker is a
40-item inventory designed to measure the FFM's five different personality dimensions
(Neuroticism, Openness to Experience, Extraversion, Conscientiousness, and
Agreeableness). Each item is an adjective (e.g., Bold, Energetic, Quiet, Systematic,
Warm) that corresponds to one of the five FFM personality dimensions. Respondents are asked to rate on a scale of 1-9 how accurately the trait describes them. Responses 1-4 represent a range between Extremely Inaccurate to Slightly Inaccurate, a response of 5 corresponds to “?”, whereas responses 6-9 represent a range between Slightly Accurate to
Extremely Accurate. Thus, higher scores represent higher accuracy of the trait description. Sum scores for each of the five personality traits were analyzed for this study.
Rosenberg Self-Esteem Scale – Bachman Revision (RSE-BR). The Rosenberg
Self-Esteem Scale is the most widely used measure to assess self-esteem (Ranzijn,
Keeves, Luszcz, & Feather, 1998). It is a self-report 10-item Guttmann scale measuring level agreement with various statements; participants have the choice between four 30 answers Strongly Agree, Agree, Disagree, and Strongly Disagree. The measure is unidimensional and scores range from 0-30, with higher scores indicating high self- esteem (Rosenberg, 1965). The Rosenberg Self-Esteem scale was originally developed with a sample of high school juniors and seniors. Though its validity has been demonstrated with a wide variety of samples and age groups, it is difficult to find its psychometric properties for a sample of older adults.
The RSE-BR was developed in 1970 (Bachman, 1970) and modified four items from the RSE, thus converting it into a multidimensional scale with two factors, positive self-regard (RSE-BR PR) and usefulness/competence (RSE-BR UC; Ranzijn et al.,
1998). Examples of items include: I am a useful person to have around, I feel that I do not have much to be proud of. Though not much emphasis has been placed on usefulness/competence in relation to self-esteem, this dimension is quite salient for use with geriatric populations. The study by Ranzijn et al. confirms the validity of the RSE-
BR with a population of older adults, and endorses the multidimensional aspect of the scale for use with older adults. As with the RSE, higher scores on the RSE-BR reflect global self-esteem and higher scores on the subscales reflect a higher sense of positive self-regard and usefulness/competence. Total RSE-B sum scores, in addition to the subscale scores were analyzed for this study.
Geriatric Depression Scale Short Form (GDS-SF). The GDS-SF is a 15-item self-report measure with yes/no answers (Yesavage et al., 1982; Yesavage & Sheikh,
1986). The instrument was designed for use among older adults with diverse levels of physical and cognitive health and can be used in a variety of settings (e.g., community, acute care, long-term care). The measure has demonstrated strong reliability and validity, 31 and has been tested against the Hamilton Rating Scale for Depression and the Zung Self-
Rating Depression Scale (Yesavage et al., 1982; Yesavage & Sheikh, 1986). Scores can range from 0 to 15, with scores of 0-4 considered normal, 5-8 indicative of mild depression, 9-11 indicative of moderate depression, and 12-15 suggesting severe depression. Depression scores were not directly analyzed in this study, however recognizing that depression is a known predictor of suicide, GDS-SF scores are likely to covary with PB and TB. In order to account for this, GDS-SF sum scores were included in a stepwise linear regression to control for depression while measuring the predictive ability of self-esteem on PB and TB.
Demographic information. Basic demographic information was collected from participants. This included age, gender, race/ethnicity, relationship status, living arrangements (e.g., alone, with spouse/significant other, with children, non-relatives), income, and education. CHAPTER 3
RESULTS
Descriptive Statistics
Means, standard deviations, and ranges were calculated for each of the study’s main variables; results are present in Table 2. In general, total GSIS mean scores and scores on the suicidal ideation scale showed moderate levels of suicidal ideation. The mean level of depression in the group suggested a normal level of depressive symptoms.
INQ means suggested low levels of thwarted belongingness and perceived burdensomeness in the sample. Self-esteem means suggested high levels of general self- esteem as well as high levels of the positive self-regard and the usefulness/competence components of self-esteem. Means for the FFM personality dimensions suggested that the majority of the sample expressed traits of Agreeableness and Conscientiousness followed by Openness to Experience and Extraversion; Neuroticism was the least expressed trait.
CATI T-scores revealed that the majority of the sample did not have clinically significant levels of PD traits. In fact, for all traits except Schizoid, only 0-3% of the sample had T- scores above 65, whereas 16% of the sample had clinically significant expressions of schizoid traits. Furthermore, CATI data suggested that 10% of the sample had clinically significant neuropsychological dysfunction.
33
Table 2
Psychometric Properties of the Major Study Variables
Range Variable M SD α Potential Actual GSIS Total GSIS 47.47 17.25 .95 31–155 31–114 Suicide Ideation 12.97 5.29 .90 10–50 10–33 Loss of Worth 11.87 4.99 .83 7–35 7–27 Death Ideation 8.66 3.64 .76 5–25 5–21 Meaning in Life 12.81 5.27 .91 8–40 8–39
INQ Total 28.18 16.40 .94 15-105 15–105 Perceived 8.30 5.63 .92 7-42 6–42 Burdensomeness Thwarted Belongingness 19.87 12.04 .95 9-63 9–63
GDS-SF 2.06 3.08 .88 0–15 0-14 RSE-BR Total RSE-BR 42.31 4.90 .84 10-50 19-47 Positive Self Regard 18.67 2.37 .90 4-20 7-20 Usefulness/Competence 23.65 2.79 .58 6-30 12-27
BFMM Extraversion 46.53 12.68 .84 8-72 16-68 Agreeableness 61.91 8.27 .85 8-72 21-72 Conscientiousness 58.62 8.76 .60 8-72 35-72 Neuroticism 25.15 9.52 .75 8-72 8-48 Openness to Experience 54.20 10.29 .85 8-72 21-72
CATI (T-scores) Antisocial 45.03 6.00 .64 20-80 32.75–71.96 Avoidant 45.15 9.34 .86 20-80 28.24–75.97 Borderline 42.06 6.84 .77 20-80 32.32–69.47 Dependent 42.48 9.09 .85 20-80 28.15–80.11 Depressive 40.88 9.95 .87 20-80 31.58–75.74 Histrionic 43.69 8.07 .70 20-80 26.66–67.05 Narcissistic 40.19 7.06 .72 20-80 25.00–60.41 Obsessive-Compulsive 41.85 9.72 .80 20-80 20.79–74.18 Paranoid 40.88 8.79 .80 20-80 25.47–62.57 Passive-Aggressive 42.89 8.86 .76 20-80 26.47–71.64 Sadistic 42.45 6.29 .51 20-80 30.48–64.13 Schizotypal 43.27 9.59 .78 20-80 26.09–68.73 Schizoid 51.61 10.92 .75 20-80 32.47–80.81 Self-Defeating 45.42 7.86 .57 20-80 25.81–66.76 Neuropsychological 47.84 11.31 .89 20-80 31.57–83.59 dysfunction 34
Reliability Measures
To ensure that our measures accurately reflected the constructs they were designed to measure, we calculated Cronbach’s α for the subscales of each measure and the measures’ total scores when applicable (see Table 2). We qualified our α scores according to George and Mallery’s (2003) cut-off points of ≥ 0.9 Excellent, 0.9 > α ≥ 0.8
Good, 0.8 > α ≥ 0.7 Acceptable, 0.7 > α ≥ 0.6 Questionable, 0.6 > α ≥ 0.5 Poor, and 0.5
> α Unacceptable. Considering that Cronbach’s α can be negatively influence by factors other than a measure’s true reliability (e.g., number of items, multidimensionality;
Cortina, 1993; Cronbach, 1951); scales in this study that demonstrate lower reliability were not eliminated from our analysis, but interpretations based on these scales were made cautiously.
The GSIS demonstrated excellent overall reliability and three of the four subscales (Suicide Ideation, Loss of Personal and Social Worth, and Meaning in Life) demonstrated good reliability; the Death Ideation subscale demonstrated acceptable reliability. The total INQ and both subscales demonstrated excellent reliability and the
GDS-SF demonstrated good reliability. Overall, the RSE-BR demonstrated good reliability, with excellent reliability for the positive self-regard subscale. However, the usefulness/competence subscale demonstrated poor reliability. Three of the five BFMM personality subscales demonstrated good reliability. The Neuroticism subscale demonstrated acceptable reliability, whereas the Conscientiousness subscale demonstrated questionable reliability. Five of the 14 CATI PD subscales demonstrated good reliability (Avoidant, Dependent, Depressive, Obsessive-Compulsive, and
Paranoid); six of the 14 CATI PD subscales demonstrated acceptable reliability 35
(Borderline, Histrionic, Narcissistic, Passive-Aggressive, Schizotypal, and Schizoid); the
Antisocial subscale demonstrated questionable reliability; whereas the Sadistic and Self-
Defeating subscales demonstrated poor reliability. The CATI neuropsychological dysfunctional scale demonstrated good/excellent reliability. In summary, the majority of the measures and subscales used in this study demonstrated acceptable to excellent reliability and none were deemed to have unacceptable reliability. The usefulness/competence subscale of self-esteem, the Conscientiousness subscale of the
BFMM, and the CATI’s Antisocial, Sadistic, and Self-Defeating subscales all demonstrated poor to questionable reliability. This suggest that results based on these subscales may not be reliable, thus any interpretation made using these results must be done so with caution.
Statistical Analyses
Hypothesis 1. A series of bivariate correlations was conducted to explore relationships between FFM dimensions, TB, PB, and SI. Results revealed significant correlations between all five FFM dimensions, TB, PB, and SI (see Table 3). The overall pattern of effect sizes suggests that the FFM personality dimensions all had their highest correlation coefficient with SI, their second highest with TB, and their lowest with PB.
Regarding TB, Agreeableness, Extraversion, Neuroticism, and Conscientiousness all demonstrated a medium effect on TB, whereas Openness to Experience had a small, but nearly medium effect on TB. As predicted, TB was positively correlated with
Neuroticism (r = .39, p < .001), and negatively with Openness to Experience (r = -.29, p
= .003), Conscientiousness (r = -.34, p = .001), Extraversion (r = -.39, p < .001), and
Agreeableness (r = -.44, p < .001). 36
Table 3
Pearson Correlations Between Personality Traits with Suicide Ideation, Perceived Burdensomeness, and Thwarted Belongingness
GSIS INQ INQ Total TB PB FFM Dimensions Extraversion -.55*** -.39*** -.31*** Agreeableness -.53*** -.44*** -.35*** Conscientiousness -.48*** -.34*** -.26*** Neuroticism *.60*** *.39*** *.32*** Openness to Experience -.46*** -.29*** -.30**
CATI Traits Antisocial .14*** .08*** .06*** Avoidant .64*** .58*** .39*** Borderline .59*** .41*** .39*** Dependent .58*** .43*** .39*** Depressive .77*** .62*** .51*** Histrionic -.09*** -.11*** .01*** Narcissistic .24** .36*** .20*** Obsessive-Compulsive .57*** .52*** .36*** Paranoid .60*** .53*** .32*** Passive-Aggressive .65*** .54*** .41*** Sadistic .17*** .12*** -.01*** Self-Defeating .56*** .46*** .29*** Schizotypal .65*** .61*** .39*** Schizoid .66*** .57*** .41*** Note. *p < .05, ** p < .01, *** p ≤ .001
Regarding PB, Agreeableness, Extraversion, Neuroticism, and Openness to
Experience had a medium effect on PB, whereas Conscientiousness has a small effect on
PB. As predicted, PB was positively correlated with Neuroticism (r = .31, p = .001), and
negatively with Openness to Experience (r = -.30, p = .002), Conscientiousness (r = -.26,
p = .008), Extraversion (r = -.31, p = .001), and Agreeableness (r = -.35, p < .001).
Regarding SI, Neuroticism, Extraversion, and Agreeableness demonstrated large
effects on SI, whereas Conscientiousness and Openness to Experience demonstrated 37 medium effects that were approaching a large effect size. Though not predicted in the hypothesis, SI was also positively correlated with Neuroticism (r = .60, p < .001), and negatively with Openness to Experience (r = -.46, p < .003), Conscientiousness (r = -.48, p < .001), Extraversion (r = -.55, p < .001), and Agreeableness (r = -.53, p < .001).
Hypothesis 2. A series of bivariate correlations were conducted between the
CATI personality disorder traits, SI, TB, and PB (see Table 3). Overall, SI, TB, and PB each had significant correlations with 11 of the 14 PD traits. Antisocial, Histrionic, and
Sadistic were the three PD traits that did not demonstrate a significant correlation with either SI, TB, or PB.
Regarding TB, seven of the 14 PD traits (Avoidant, Depressive, Obsessive-
Compulsive, Paranoid, Passive-Aggressive, Schizotypal, and Schizoid) had large significant relationships with TB, whereas four of the 14 PD traits (Borderline,
Dependent, Narcissistic, and Self-Defeating) had a moderate significant relationship with
TB. We successfully predicted that TB would correlate significantly with Avoidant (r =
.58, p < .001), Dependent (r = .43, p < .001), Depressive (r = .62, p < .001), Obsessive-
Compulsive (r = .52, p < .001), Paranoid (r = .53, p < .001), Passive-Aggressive (r = .54, p < .001), Schizoid (r = .57, p < .001), Schizotypal (r = .61, p < .001), and Self-Defeating
(r = .46, p < .001) PD traits. We also predicted that Antisocial, Sadistic, and Histrionic traits would have a significant positive correlation with TB and though they were positively correlated, they failed to achieve statistical significance. Though not included in our hypothesis, a significant relationship was also discovered between TB and
Borderline (r = .41, p < .001) and Narcissistic (r = .36, p < .001) PD traits.
38
Regarding PB, only one (Depressive) of the 14 PD traits had large significant relationships with PB, eight (Avoidant, Borderline, Dependent, Obsessive-Compulsive,
Paranoid, Passive-Aggressive, Schizotypal, and Schizoid) of the 14 PD traits demonstrated a moderate effect on PB, whereas two of the 14 PD traits (Narcissistic, and
Self-Defeating) demonstrated a small effect on PB. We successfully predicted that PB would correlate significantly with Avoidant (r = .39, p < .001), Dependent (r = .39, p <
.001), Depressive (r = .51, p < .001), Obsessive-Compulsive (r = .36, p < .001), Paranoid
(r = .32, p = .001), Passive-Aggressive (r = .41, p < .001), Schizoid (r = .41, p < .001),
Schizotypal (r = .39, p < .001), and Self-Defeating (r = .29, p = .003) PD traits. We also predicted that Antisocial, and Sadistic traits would have a significant positive correlation with PB and though they were positively correlated, they failed to achieve statistical significance. Though not included in our hypothesis, a significant relationship was also discovered between TB and Borderline (r = .39, p < .001) and Narcissistic (r = .20, p =
.043) PD traits.
In our hypothesis we did not predict which specific PD traits would correlated significantly with SI, however 11 of the 14 PDs achieved statistical significance, all in the positive direction. Ten of the 14 PD traits demonstrated large effect sizes on SI.
Specifically, large significant effects were found between SI with Avoidant (r = .64, p <
.001), Borderline (r = .59, p < .001), Dependent (r = .58, p < .001), Depressive (r = .77, p
< .001), Obsessive-Compulsive (r = .57, p < .001), Paranoid (r = .60, p < .001), Passive-
Aggressive (r = .65, p < .001), Self-Defeating (r = .56, p < .001), Schizotypal (r = .65, p
< .001), and Schizoid (r = .66, p < .001) PD traits; whereas there was a small significant relationship between SI and Narcissistic PD trait (r = .24, p = .016). 39
The pattern of correlation coefficients between PD traits, SI, TB, and PB that emerged was similar to the pattern observed with the FFM dimensions, indicating that all
PD traits had their highest correlation coefficient with SI, then TB, and PB. This was true with the exception of Narcissistic PD traits, which had its strongest correlation with TB, then SI, and finally PB.
Hypothesis 3. A series of bivariate correlations were conducted to determine the relationships between self-esteem (RSE-BR total, RSE-BR PR, and RSE-BR UC), GSIS-
LPSW, TB, and PB. Though predictions about the relationship between SI and self-esteem were not made in our hypothesis, we have included the data in our table as reference (see
Table 4). All variables of interest demonstrated moderate to large significant relationships with TB, and PB. As with results from the previous two hypotheses, all variables demonstrated their highest correlation coefficient with SI, then TB, and then PB.
Table 4
Pearson Correlations Between Self-Esteem with Suicidal Ideation, Perceived Burdensomeness, and Thwarted Belongingness
GSIS INQ INQ Total TB PB
RSE-BR Total -.74*** -.63*** -.53*** RSE-BR PR -.73*** -.59*** -.56*** RSE-BR UC -.68*** -.61*** -.46** GSIS LPSW .91*** .60*** .54*** GSIS SI .93*** .57*** .57*** GSIS DI .78*** .38*** .37*** GSIS PML *.89*** *.56*** *.59*** Note. RSE-BR SR = Rosenberg Self-Esteem Scale Bachman Revision Positive Self Regard; RSE-BR UC = Rosenberg Self-Esteem Scale Bachman Revision Usefulness/Competence; GSIS LPSW = Geriatric Suicidal Ideation Scale – Loss of Personal and Social Worth; GSIS SI = Geriatric Suicidal Ideation Scale – Suicide Ideation; GSIS DI = Geriatric Suicidal Ideation Scale – Death Ideation; GSIS PML = Geriatric Suicidal Ideation Scale – Perceived Meaning in Life. ** p < .01, *** p ≤ .001
40
Regarding TB, as predicted, it had a significant positive relationship with GSIS-
LPSW (r = .60, p < .001) and a significant negative relationship with RSE-BR total (r = -
.63, p < .001), RSE-BR SR (r = -.59, p < .001) and RSE-BR UC (r = -.61, p < .001).
Thus, TB demonstrated a large significant relationship with total self-esteem and its two subscales, as well as with the GSIS loss of personal and social worth subscale. We also predicted that TB would correlate more strongly with the positive self-regard component of self-esteem compared to the usefulness/competence component of self-esteem; this hypothesis was not supported. Despite the fact that TB correlated more strongly with
RSE-BR UC (r = -.61, p < .001) than with RSE-BR PR (r = -.59, p < .001), by calculating a z statistic for these correlation coefficients, we determined that this difference was not statistically significant (z = -.28, p = .389).
Regarding PB, as predicted, there was a significant positive relationship with
GSIS-LPSW (r = .54, p < .001) and a significant negative relationship with RSE-BR total
(r = -.53, p < .001), RSE-BR PR (r = -.56, p < .001) and RSE-BR UC (r = -.46 p < .005).
Thus, PB demonstrated a large significant relationship with total self-esteem, the positive self-regard component of self-esteem and with GSIS-LPSW, however, PB demonstrated a moderate significant relationship with RSE-BR UC. We also predicted that PB would correlate more strongly with RSE-BR UC than with RSE-BR SR. This hypothesis was not supported; PB demonstrated a large negative correlation with RSE-BR SR (r = -.56, p
< .001) and a medium correlation with the RSE-BR UC (r = -.46, p < .001). We calculated a z-statistic to quantify the difference between these correlation coefficients, and determined that the usefulness/competence component had a significantly smaller relationship with PB than the positive self-regard component (z = 1.92, p = .027). 41
Hypothesis 4. Separate multiple linear regressions were conducted to regress the two different personality models onto three separate outcome variables: SI, TB, and PB.
For the FFM regression, previous research demonstrated that Neuroticism and
Extraversion had strong relationships with TB and PB (Cramer et al., 2012; DeShong et al., 2015), thus they were entered into the model first, followed by the three remaining personality traits (Openness to Experience, Agreeableness, and Conscientiousness).
Diagnostic statistics and graphs revealed that the SI and FFM data was normally distributed, that there was independence of errors (Durbin-Watson value = 2.3), and that the data were homoscedastic. As such the FFM and SI data were analyzed without corrections. Diagnostic statistics and graphs revealed that the FFM, TB, and PB data demonstrated non-normal distribution and some heteroscedasticity, but met the assumption of independent errors (Durbin-Watson value = 2.0 – TB, Durbin-Watson value = 1.95 – PB). To correct for the non-normal distribution of TB and PB, confident intervals were generated using bias corrected and accelerated bootstrapping with 1000 replications. Bootstrapping is a statistical technique that is robust to violations of parametric assumptions. This technique treats the dataset as a population from which it takes multiple random samples—usually 1000 to 2000 different samples. The process then calculates the parameter of interest for each sample and synthesizes these results in the form of confidence intervals for the parameter (Field, 2013). As such, this process is robust to assumption violations with only a small loss in predictive power (Davidson &
MacKinnon, 1996).
FFM. Regarding SI, the FFM model was significantly predictive F(5,96) = 20.22
(p < .001) and accounted for 51.3% of the variance in SI (R2 = .51, R2 adjusted = .49). Of 42 the five personality dimensions only two were statistically significant predictors at the independent level; Neuroticism (β = .58, p = .002), 95% CI [ 0.26, 0.90] positively predicted SI and Extraversion (β = -.35, p = .016), 95% CI [-0.67, -0.05] negatively predicted SI (see Table 5). Therefore, as Neuroticism increases so does SI, whereas when Extraversion increases SI decreases.
Table 5
Linear Regression of Five Factor Model Dimensions on Suicidal Ideation
Variable B SE B t 95% CI p Neuroticism .58 .16 3.58 [0.26, 0.90] .001 Extraversion -.35 .12 -2.87 [-0.59, -0.11] .005 Agreeableness -.30 .19 -1.53 [-0.68, 0.09] .130 Conscientiousness -.30 .18 -1.69 [-0.67, 0.05] .094 Openness to -.11 .16 -0.68 [-0.43, 0.21] .500 Experience R2 .51 Ṝ 2 .49 F 20.22 .000
Regarding PB, the FFM model was significantly predictive F(5,96) = 4.06 (p =
.002) and accounted for 17.5% of the variance in PB. However, none of the specific FFM dimensions independently predicted PB (see Table 6).
Table 6
Bootstrapped Linear Regression of Five Factor Model Dimensions on Perceived Burdensomeness
Variable B SE B t BCa 95% CI p Neuroticism .07 .05 -0.06 [-0.02, 0.18] .180 Extraversion -.05 .04 -0.16 [-0.14, 0.04] .268 Agreeableness -.12 .14 -0.28 [-0.38, 0.19] .510 Conscientiousness -.03 .08 -0.18 [-0.14, 0.14] .752 Openness to -.05 .07 -0.18 [-0.20, 0.07] .538 Experience R2 .18 Ṝ 2 .13 F 4.07 .002 43
The FFM model was also significantly predictive of TB F(5,96) = 6.94 (p < .001) and account for 26.5% of the variance. With bootstrapping, Extraversion was a significant independent negative predictor of TB (β = -.18, p = .043), 95% BCa CI [-0.36,
0.001] (see Table 7). Given the negative relationship, as Extraversion increases TB is expected to decrease.
Table 7
Bootstrapped Linear Regression of Five Factor Model Dimensions on Thwarted Belongingness
Variable B SE B t BCa 95% CI p Neuroticism .18 .13 1.29 [-0.06, 0.45] .160 Extraversion -.18 .09 -1.73 [-0.36, -0.00] .043 Agreeableness -.35 .17 -2.10 [-0.66, 0.01] .052 Conscientiousness -.17 .15 -1.11 [-0.43, 0.11] .252 Openness to .05 .13 0.33 [-0.26, 0.34] .753 Experience R2 .27 Ṝ 2 .28 F 6.94 .000
As an additional analysis, the FFM dimensions were also regressed onto to the total INQ score. The FFM model was significantly predictive of the total INQ score
F(5,96) = 7.08 (p < .001) and accounted for 26.9% of the variance. Agreeableness was a significant negative predictor of total INQ scores (β = -.47, p = .042), 95% CI [-0.92, -
0.02] (see Table 8). This relationship means that as Agreeableness increases, total INQ scores will decrease.
PD traits. In the goal of obtaining a model for the PD traits with a good statistical fit without including too many predictors, only the 10 current DSM-5 PDs were entered as predictor variables. The PD traits were grouped according to the three PD cluster profiles. The order of the predictor variables was based on past research that identified 44
Table 8
Linear Regression of Five Factor Model Dimensions on Total INQ Score
Variable B SE B t 95% CI p Neuroticism .25 .19 1.33 [-0.12, 0.63] .184 Extraversion -.23 .14 -1.64 [-0.52, 0.05] .103 Agreeableness -.46 .47 -2.07 [-0.92, -0.02] .042 Conscientiousness -.20 .21 -0.94 [-0.62, 0.22] .351 Openness to -.01 .19 -0.00 [-0.37, 0.37] .984 Experience R2 .27 Ṝ 2 .23 F 7.08 .000 the predictive ability of the different PDs on SI (Eades et al., 2015) and were grouped by
PD clusters. Clusters that contained PDs with the strongest predictive ability were entered first. Thus, the cluster A PDs (i.e., Paranoid, Schizoid, Schizotypal) were entered first in the list, then the cluster C PDs (i.e., Dependent, Avoidant, Obsessive-Compulsive) and finally the cluster B PDs (i.e., antisocial, Borderline, Histrionic, Narcissistic). The diagnostic statistics and charts demonstrated that the PD and SI data was normally distributed, that it was heteroscedastic and that there was a slight negative correlation in the residuals (Durbin-Watson value = 2.4). Because of the normal distribution of data, the
PD and SI data was interpreted without correction. The PD and TB data did not meet the assumption of normality, or the assumption of homoscedasticity, but it did demonstrate independence of error (Durbin Watson value = 1.97). To correct for the non-normal distribution of the data, confidence intervals for the regression model were generated by completing bias corrected bootstrapping with 1000 replications. The PD and TB data appeared to have a loosely normal distribution, some heteroscedasticity, and had a small positive correlation within its residuals (Durbin-Watson value = 1.76). Because the 45 distribution of the PD and TB data was not undoubtedly normal, confidence intervals for the regression model were generated by completing bias corrected and accelerated bootstrapping with 1000 replications.
Regarding SI, the model using the 10 DSM-5 PDs was significantly predictive
F(10,91) = 20.86 (p < .001) and accounted for 69.6% of the variance (R2 = .70, R2 adjusted = .66). Without bootstrapping, four of the 10 PDs were found to significantly predict SI in the positive direction: Paranoid PD (β = .44, p = .04), 95% CI [0.01, 0.86),
Schizoid (β = .70, p < .001), 95% CI [0.35, 1.04], Borderline (β = .86, p = .001) 95% CI
[0.36, 1.37], and Dependent (β = .81, p < .001), 95% CI [0.37, 1.25] (see Table 9).
Table 9
Linear Regression of Personality Disorder Traits on Suicidal Ideation
Variable B SE B t 95% CI p Paranoid .44 .21 2.05 [0.01, 0.87] .043 Schizotypal .16 .10 0.82 [-0.23, 0.55] .416 Schizoid .70 .18 3.98 [0.35, 1.04] .000 Avoidant -.32 .26 -1.23 [-0.83, 0.20] .226 Dependent .81 .22 3.69 [0.37, 1.25] .000 Obsessive-Compulsive -.27 .20 -1.31 [-0.67, 0.14] .193 Antisocial -.23 .19 -1.20 [-0.62, 0.15] .232 Borderline .86 .25 3.40 [0.34, 1.37] .001 Histrionic -.15 .24 -0.64 [-0.62, 0.32] .526 Narcissistic -.33 .23 -1.46 [-0.78, 0.12] .148 R2 .70 Ṝ 2 .66 F 20.86 .000
Regarding PB, the model with the 10 PD traits was significantly predictive F(10,
91) = 3.75 (p < .001) and accounted for 29.2% of the variance (R2 = .29, R2 adjusted =
.21). With bootstrapping, Schizoid PD (β = .24, p = .041), 95% BCa CI [0.08, 0.35] was the only significant independent predictor of PB (see Table 10). In this positive 46 relationship, as Schizoid traits increase so does PB. Regarding TB, the 10 PD model was also significantly predictive F(10, 91) = 7.84 (p < .001) and accounted for 46.3% of the variance in TB (R2 = .46, R2 adjusted = .40), however no one PD was independently predictive of TB (see Table 11).
Table 10
Bootstrapped Linear Regression of Personality Disorder Traits on Perceived Burdensomeness
Variable B SE B t BCa 95% CI p Paranoid -.05 .15 -0.44 [-0.38, 0.18] .778 Schizotypal .01 .08 0.04 [-0.15, 0.21] .956 Schizoid .21 .08 2.42 [0.08, 0.35] .041 Avoidant -.09 .11 -0.68 [-0.31, 0.23] .445 Dependent .17 .19 1.54 [-0.26, 0.44] .451 Obsessive-Compulsive -.04 .13 -0.38 [-0.25, 0.23] .786 Antisocial -.07 .11 -0.74 [-0.37, 0.10] .555 Borderline .24 .14 1.90 [-0.05, 0.48] .118 Histrionic -.02 .13 -0.16 [-0.25, 0.28] .889 Narcissistic .04 .07 0.37 [-0.08, 0.20] .568 R2 .29 Ṝ 2 .21 F 3.75 .000
Table 11
Bootstrapped Linear Regression of Personality Disorder Traits on Thwarted Belongingness
Variable B SE B t BCa 95% CI p Paranoid .00 .28 0.01 [-0.53, 0.43] .088 Schizotypal .33 .18 1.83 [-0.05, 0.71] .999 Schizoid .22 .15 1.33 [-0.06, 0.49] .074 Avoidant .08 .24 0.31 [-0.40, 0.57] .183 Dependent .13 .27 0.65 [-0.39, 0.63] .724 Obsessive-Compulsive -.08 .22 -0.41 [-0.54, 0.42] .620 Antisocial -.22 .19 -1.21 [-0.58, 0.15] .723 Borderline .38 .28 1.61 [-0.14, 1.01] .242 Histrionic -.18 .25 -0.81 [-0.66, 0.26] .179 Narcissistic .25 .19 1.20 [-0.11, 0.65] .494 R2 .46 Ṝ 2 .40 F 7.84 .000 47
As an additional analysis, the PD traits were also regressed onto to the total
INQ score. The PD model was significantly predictive of total INQ scores F(10,91)
= 7.41 (p < .001) and accounted for 26.9% of the variance. None of the PD traits were independently predictive of the total INQ score (see Table 12).
Table 12
Linear Regression of Personality Disorder Traits on Total INQ Score
Variable B SE B t 95% CI p Paranoid -.05 .27 -1.07 [-0.59, 0.50] .867 Schizotypal .34 .25 1.35 [-0.16, 0.83] .182 Schizoid .43 .22 1.91 [-0.20, 0.87] .060 Avoidant -.01 .33 -1.04 [-0.67, 0.65] .971 Dependent .30 .28 1.07 [-0.26, 0.86] .288 Obsessive- -.12 .26 -0.44 [-0.63, 0.40] .659 Compulsive Antisocial -.29 .25 -1.16 [-0.78, 0.20] .248 Borderline .62 .33 1.91 [-0.03, 1.27] .060 Histrionic -.20 .30 -0.65 [-0.79, 0.40] .518 Narcissistic .29 .29 1.01 [-0.28, 0.87] .315 R2 .45 Ṝ 2 .39 F 7.42 .000
Hypothesis 5. To explore the predictive ability of self-esteem on TB and PB while controlling for depression, a hierarchical multiple linear regression was conducted.
The total score for the Geriatric Depression Scale was entered as the first block in the model followed by GSIS-PSW, RSEBR-SR, and RSEBR-UC. This model was run twice, once for TB and once for PB as the outcome variable. With depression scores held constant, the positive self-regard component of self-esteem was a negative significant predictor of PB (β = -0.70, p = .042), 95% CI [-1.38, -0.03], with a partial correlation score of -.24 (p = 0.18). With depression scores held constant, the usefulness/competence 48 component of self-esteem was a significant negative predictor of TB (β = -1.25, p =
.023), 95% CI [-2.32, -0.18], with a partial correlation score of -.37 (p < .001). Loss of social and personal self-worth as identified in the GSIS did not significantly predict either
TB or PB. Thus, as RSE-BR SR increases PB decreases, whereas when RSE-BR UC increases TB decreases.
Though not part of our hypotheses, in order to gain additional information about self-esteem, two additional regressions were performed. First, RSE-BR SR and RSE-BR
UC were regressed onto the total GSIS score (SI) and then again onto the total INQ score.
The self-esteem model was significantly predictive of SI F(2,99) = 61.85 (p < .001) and accounted for 55.5% of the variance. Both RSE-BR SR (β = -3.63, p < .001), 95% CI [-
5.24, -2.02] and RSE-BR UC (β = -1.75, p = .013), 95% CI [-3.12, -0.39] were significant independent negative predictors of SI. The self-esteem model was also significantly predictive of the total INQ score F(2,99) = 36.17 (p < .001) and accounted for 42.2% of the variance. Both RSE-BR SR (β = -2.80, p = .002), 95% CI [-4.55, -1.05] and RSE-BR UC (β = -1.64, p = .030), 95% CI [-3.13, -0.16] were significant independent negative predictors of total INQ scores. Thus, as RSE-BR SR and RSE-BR
UC increase, both SI and the total INQ score will decrease.
Exploratory Analyses
Mediation of personality and suicidal ideation. To prepare the data for a mediation analysis exploring the impact of TB and PB on the relationship between PDs and SI, a single outcome score was created to represent a participant’s total level of personality disturbance. This was done by adding the T-scores for all 10 current DSM-5
PDs and taking the average. This average T-score was then used as the predictor variable 49 in the model, TB and PB scores were both included in the model as mediator variables, and SI was used as the outcome variable. The analysis was completed using the
PROCESS packet for SPSS (Hayes, 2013). Together, TB and PB exerted a significant indirect effect on the relationship between PDs and SI, b = 0.73 95% BCa CI [0.12, 1.97], the standardized indirect effect was small in size b = .22, 95% BCa CI [0.47, 0.57]. This signifies that the relationship between PDs and SI is mediated by TB and PB. However, independently neither TP or PB were found to have a significant indirect effect on the relationship between PDs and SI.
As an additional exploratory mediation analysis, the two subscales of the RSE-B scale were used as mediators for the relationship between PDs and SI. Combined, RSE-
BR SR and RSE-BR UC had a significant indirect effect on the relationship between PDs and SI, b = 1.18, BCa 95% CI [ 0.56, 1.92]. The total standardized effect of self-esteem on PDs and SI was moderate in size b = .36, BCa 95% CI [0.19, 0.54]. This suggests that self-esteem significantly mediates the relationship between PDs and SI. Analyses also demonstrated that RSE-BP SR had a significant indirect effect on the relationship between PDs and SI, b = 0.82, BCa 95% CI [0.07, 1.69]. The standardized effect was small in size b = 0.25, BCa 95% CI [0.02, 0.46]. This suggests that positive self-regard exerts more influence over the relationship between PD and SI than one’s perception of usefulness/competence.
Cognition. A series of multiple linear regressions were conducted to explore the relationship between neuropsychological factors, SI, TB, and PB. Specifically, the neuropsychological dysfunction scale (CATI NPD), the executive function decision difficulty scale (CATI EXDD), and the executive function planning problems scale 50
(CATI EXPP) from the CATI were regressed onto SI, TP and PB separately. Diagnostic statistics revealed that the data in the SI, TB, and PB model did not meet the assumption of normality, therefore bias corrected and accelerated bootstrapping (BCa) was conducted with 1000 replications.
Regarding SI, results revealed a strong positive correlation between SI and CATI NPD (r
= .54, p < .001), as well as SI and CATI EXDD (r = .47, p < .001); SI was not significantly correlated with CATI EXPP (see Table 13). In a linear regression model using the three cognitive measures, the model was significantly predictive of SI F (3, 98)
= 18.50, (p < .001) and accounted for 36.2% of the variance in SI (see Table 14).
Table 13
Pearson Correlations Between CATI Neuropsychological Variables with Suicidal Ideation, Perceived Burdensomeness, and Thwarted Belongingness
GSIS INQ INQ Total TB PB
NPD .57*** .46*** .31*** EXDD .47*** .32*** .33*** EXPP *.10 *** .21** .22** Note. NPD = Neuropsychological Deficit Scale; EXDD = Executive Function Decision Difficulty; EXPP = Executive Function Planning Problems * p ≤ .05, *** p ≤ .001
Table 14
Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Suicidal Ideation
CATI Variable B SE B t BCa 95% CI p NPD .63 .17 4.60 [0.28, 0.98] .003 EXDD .51 .24 3.22 [0.03, 0.89] .047 EXPP .01 .22 .03 [-0.43, 0.40] .976 R2 .36 Ṝ 2 .34 F 18.50 .000 Note. NPD = Neuropsychological Deficit Scale; EXDD = Executive Function Decision Difficulty; EXPP = Executive Function Planning Problems 51
At the individual level, total CATI NPD (β = .63, p = .003), BCa 95% CI [0.28, 0.98] and
CATI EXDD (β = .51, p = .047), BCa 95% CI [0.03, 0.89] were both significant predictors of SI. Both of these relationships were positive, thus as overall neuropsychological dysfunction and executive decision making difficulties increase, so does SI.
Regarding PB, analyses revealed small to moderate positive correlations between
PB, CATI NPD (r = .31, p = .001), CATI EXDD (r = .33, p < .001), and CATI EXPP (r
= .22, p = .013). The linear regression model using these three variables was significantly predictive of PB F(3, 98) = 6.71, (p < .001) and accounted for 17% of the variance (R2 =
.170). After BCa bootstrapping, none of the three predictor variables were individually able to predict PB (see Table 15).
Table 15
Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Perceived Burdensomeness
CATI Variable B SE B t BCa 95% CI p NPD .11 .06 2.17 [-0.01, 0.24] .085 EXDD .11 .13 1.90 [-0.15, 0.31] .486 EXPP .13 .12 1.66 [-0.06, 0.38] .329 R2 .17 Ṝ 2 .15 F 6.71 .000 Note. NPD = Neuropsychological Deficit Scale; EXDD = Executive Function Decision Difficulty; EXPP = Executive Function Planning Problems
Regarding TB, analyses revealed small to moderate positive correlations between
TB, CATI NPD (r = .46, p < .001), CATI EXDD (r = .33, p < .001), and CATI EXPP (r
= .21, p = .016). The linear regression model using these three variables was significantly predictive of TB F(3, 98) = 11.38, (p < .001) and accounted for 25.8% of the variance (R2 52
= .258). After BCa bootstrapping, only CATI NPD was able to significantly predict TB
(β = .44, p = .001), BCa 95% CI [0.22, 0.69] (see Table 16). The positive direction of this relationship means that as neuropsychological dysfunction increases, so does TB.
Table 16
Bootstrapped Linear Regression for CATI Neuropsychological Predictors of Thwarted Belongingness
CATI Variable B SE B t BCa 95% CI p NPD .44 .10 0.53 [0.22, 0.69] .001 EXDD .13 .15 0.46 [-0.18, 0.37] .428 EXPP .30 .20 -0.14 [-0.13, 0.76] .151 R2 .26 Ṝ 2 .24 F 11.38 .000 Note. NPD = Neuropsychological Deficit Scale; EXDD = Executive Function Decision Difficulty; EXPP = Executive Function Planning Problems CHAPTER 4
DISCUSSION
Thwarted belongingness and perceived burdensomeness are two well established risk factors for suicidal ideation, however little is known about how personality and self- esteem relate to TB and PB. As such, the primary purpose of this study was to understand the relationship between TB, PB, personality, and self-esteem and how these relationships might be important to SI among older adults—a group at elevated risk for suicide. This study also offered an opportunity to reflect upon how and why personality and self-esteem may be connected TB and PB.
General
A notable theme across our results is the pattern of relationship between our independent variables of interest and our dependent variables. Specifically, results from our correlations demonstrated that all FFM personality dimensions and PD traits (with the exception of Narcissism) demonstrated their highest correlation coefficients with SI, then
TB and finally PB. Furthermore, our linear regressions demonstrated that our different personality models accounted for the most amount of variance in SI, TB, and then PB.
This pattern of results was true of our self-esteem variables as well. Though past research does not have data available for all of our independent variables and methods of analyses, data from Jahn et al. (2015) and Eades et al. (2015) also demonstrated the same pattern of correlation between PD traits, TB, and PB. As such, this might be an established pattern
54 that should be explored in future studies.
Joiner et al. (2009) and Van Orden et al. (2010) conceptualize suicidal ideation as the result of coexistent TB and PB. From this perspective, SI is essentially the consequence of adding TB and PB together. As such it is logical that independently, TB and PB would have a smaller relationship with our independent variables compared to the relationship between the same variables and SI. Thus, the pattern of results found in this study might in fact be support for the IPTS. Considering that the total INQ score is the sum of TB and PB scores, as an exploratory analysis we decided to compare the proportion of INQ and SI variance accounted for by our FFM, PD, and self-esteem variables. Our analyses revealed that the different models accounted for more of the variance in SI than in the total INQ score. This pattern is similar to the one observed in our correlations and regression models with SI, TB, and PB. Given that each model is accounting for more of the variance in SI than in the INQ, it is possible that our independent variables comprise elements that are connected to SI but not directly connected with TB and PB. This suggests that there are elements beyond TB and PB that contribute to SI. Some of these elements may include the desire to escape, levels of functioning and autonomy, psychological problems (Draper, Kõlves, De Leo, &
Snowdon, 2014; Van Orden et al., 2015) or more existential concepts such as reasons for living and meaning in life (Heisel, Neufeld, & Flett, 2016). Future research exploring the variables related to SI that fall outside the concepts of TB and PB may serve to deepen our understanding of SI and increase our ability to predict and prevent SI among older adults.
55
Personality
FFM. Costa and McCrea's (1992) Five Factor Model of personality suggests that personality is combination of five different normative personality dimensions and that all individuals exhibit different levels of these traits. With this in mind, our first hypothesis assumed that different FFM dimensions would correlate in different ways with TB, PB, and SI. As predicted, all five dimensions were significantly correlated with TB, PB, and
SI. Of the five personality dimensions, Agreeableness had the strongest relationship with
TB and PB, whereas Neuroticism had the strongest relationship with suicidal ideation.
Our results regarding SI and the FFM dimensions align with the results of previous research with older adults which demonstrate a strong positive relationship between Neuroticism and SI (Heisel & Flett, 2006; Iliceto et al., 2014; Segal et al., 2012;
Useda et al., 2004), as well as a strong negative relationship between Extraversion and SI
(Iliceto et al., 2014; Segal et al., 2012). Though previous research suggested little to no relationship between SI and Openness to Experience, Agreeableness, and
Conscientiousness, results from this study suggest a large negative correlation between each of these variables and SI. However, results from our linear regression demonstrate that Neuroticism and Extraversion were the only two dimensions that could significantly predict levels of SI. As such, our results support the results from previous studies.
Given that little personality research has been conducted within a framework of the IPTS, this study provides a rich opportunity to consider how different personality dimensions may be connected to TB and PB. The moderate negative relationship between Agreeableness and TB might be explained in consideration of loneliness. As one of TB’s latent constructs, high loneliness is likely to increase levels of TB (Van Orden et
56 al., 2010). Furthermore, it is conceivable that individuals who express high levels of
Agreeableness would have more frequent social contact, therefore reducing loneliness and TB. Furthering this idea, the results of our bootstrapped linear regression for TB and the FFM dimensions suggest that Extraversion is a significant negative predictor of TB.
Given that high levels of Extraversion might lead an individual to seek out the company of others, it is plausible that this could in turn decrease loneliness and therefore TB.
Future research could be used to explore this pathway in greater depth.
The moderate negative relationship relationship between PB and Agreeableness is not as easily explained. PB is composed of two latent variables: liability and self-hatred.
Van Orden et al. (2010) suggest that indicators of self-hatred may include low self- esteem, self-blame, shame, or agitation. One potential explanation for the relationship between PB and Agreeableness is that individuals could be more prone to self-hatred if they do not feel as though they are being agreeable. In a study of the personality correlates of self-esteem, Robins et al., (2001) observed a slightly stronger correlation between Agreeableness and self-esteem for women than for men. Their data also revealed that the relationship between self-esteem and FFM dimensions (including Agreeableness) increased among their participants over the age of 60 compared to their younger participants. Given that our sample had a higher proportion of women, and consisted only of individuals over the age of 60, there may be reason to believe that Agreeableness is connected to PB via self-esteem. Alternatively, it is possible that Agreeableness is related to PB through the liability construct. Perhaps individuals feel as though they are less of a liability to their loved ones if they believe that they are being agreeable. Though both of these explanations for the relationship between PB and Agreeableness are plausible,
57 additional research would be required to confirm this hypothesis.
Additional research exploring the relationship between FFM dimensions and PB is also warranted given that our findings diverge from recently published data
(Vanyukov, Szanto, Hallquist, Moitra, & Dombrovski, 2016). In their study, Vanyukov et al. reported that PB correlated positively with Neuroticism and negatively with
Extraversion, Openness, Agreeableness, and Conscientiousness. Though the directions of correlation in our study align with their results, the strength of the relationships were notably different. For example, in our study all five variables correlated significantly with
PB, and the strongest relationship was found between PB and Agreeableness (r = -.35).
However, in their study only 3 of the five dimensions were significantly correlated with
PB and Neuroticism had the strongest correlation (r = .52) with PB. Furthermore,
Agreeableness had the weakest correlation (r = -.13) with PB and this relationship failed to achieve significance. Though it is possible that some of these differences are due to their younger sample (42 years of age and older) and their inclusion of a more symptomatically diverse group of participants (high-lethality attempters, low-lethality attempters, depressed non-suicidal, and healthy controls) additional research would be required to explore these difference more thoroughly.
Personality disorder traits. Another way to conceptualize personality is from a clinical perspective based on the 10 personality disorders identified in the DSM-5. To be diagnosed with a PD, an individual must meet a minimum number of symptoms.
However, by conceptualizing PDs from a dimensional perspective, a PD diagnosis represents an abnormally large expression of specific personality traits. As such, it can be assumed that within a non-clinical population, individuals may still express a certain
58 level of personality disorder symptomology but that their number of symptoms do not meet the threshold for clinical significance. Thus, this study aimed to explore the relationship between different levels of PD trait-expression, PB, TB, and SI among older adults.
Of the 10 PDs, the strongest significant relationships with TB were found between Schizotypal, Avoidant, and Schizoid PD traits. All of these PDs comprise elements of social discomfort, a reduced capacity for close relationships, and interpersonal difficulty. TB comprises the latent components of loneliness and absence of reciprocal care. Given the interpersonal symptoms associated with Schizotypal, Avoidant, and Schizoid traits, it is likely they are connected to TB through both the loneliness constructs and the absence of reciprocal care construct.
PB comprises the liability and self-hatred constructs and demonstrated the strongest relationship with Schizoid followed by a tie between Schizotypal, Avoidant,
Borderline, and Dependent traits. Schizoid PD is characterized by a “pervasive pattern of detachment from social relationships” (American Psychiatric Association, 2013, p. 653) and restricted emotional expression in interpersonal settings. Considering that social relationships cause distress for individuals with Schizoid PD, it is easy to conceive that individuals with Schizoid traits would be inclined to classify themselves as a burden on others if they feel as though they are reliant on others. Consequently, it is likely that the schizoid symptomology is tapping into the liability construct of PB. Interestingly,
Schizoid traits were also strongly correlated with TB. This suggests that even though individuals with Schizoid traits seek out social isolation, this isolation may still be distressing to them. Furthermore, Avoidant, Borderline, and Dependent PDs all have
59 symptomatology related impairments in self-functioning without input or support from others. It is possible that this symptomatology is connected to the self-hatred component of PB in that, individuals with these traits may not have a strong enough sense of self to offset self-hatred. Alternatively, these individuals may also recognize (or be told that) that they are dependent on others which may be connected to the liability construct and causing them to experience higher levels of PB.
SI was also highly correlated with the same personality traits as TB and PB, which suggests that the TB and PB experienced by individuals with these personality traits could potentially be distressing enough for them to contemplate suicide. This also could be additional support for the IPTS’ central argument which states that when TB and
SI are both present, it is likely to lead to SI (Joiner, 2005; Van Orden et al., 2010).
Though personality has traditionally been considered a stable construct, there is evidence that changes in personality may slow with age but that they do not necessarily cease, especially in the case of PD traits (Cooper, Balsis, & Oltmanns, 2014; Debast et al., 2014; Mordekar & Spence, 2008; Newton-Howes, Clark, & Chanen, 2015). To better understand the relationship between PD traits, TB, PB, and SI across the life span future research may consider including a longitudinal or cross sectional component. Future studies may benefit from including a sample of individuals who were diagnosed with a
PD late in life as well as a sample of individuals whose PD’s symptoms have lessened with age. Differences between these samples may help us to better understand which PD traits are likely to decrease with age and/or which PD traits are more likely to emerge in later life. This information may have important clinical implications in the context of long-term care settings—or for family caregivers—as increased education about late-life
60
PDs may help individuals to reinterpret “difficult behavior”, thus decreasing frustration for the patient and the provider, which should help to increase the quality of care.
FFM vs personality disorder traits. Although it is difficult to statistically determine whether FFM dimensions or PD traits are better predictors of TB, PB, and SI, we can cautiously examine the percentage of variance accounted for by both models.
Overall, the PD traits model consistently accounted for more variance in our outcome predictors compared to the FFM model. For TB, the FFM model accounted for 26.5% of the variance, whereas the PD model accounted for 46.3%. For PB, the FFM model accounted for 17.5% of the variance, whereas the PD model accounted for 29.2%; and for
SI, the FFM model accounted for 51.3% of the variance, whereas the PD model accounted for 69.9% of the variance. Though it would be easy to assume that the PD model is a better statistical fit because it consistently accounts for more variance, this might be due to the fact that it included 10 predictor variables compare to only 5 in the
FFM model. Consequently, a more rigorous statistical analysis—which is outside the scope of this project—would be required to in order to explore this further and draw more meaningful conclusions.
Self-Esteem
General self-esteem. In this study we were interested in exploring the relationship between self-esteem and suicidal ideation among older adults. Given that TB and PB are established predictors of SI, we also wanted to understand if there is a relationship between TB, PB and self-esteem. Results revealed a very strong negative relationship between SI and self-esteem, and it was determined that although both constructs of self-esteem had a large significant relationship with SI, positive self-regard
61 had a stronger relationship with SI than perception of usefulness/competence. This suggests that when individuals have a positive self-perception and when they feel useful/competent they are less likely to think about suicide. This aligns with previous research with older Norwegian suicide attempters that demonstrated a link between suicide attempts, a perceived loss of achievement, and low personal value (Kjølseth et al.,
2009).
Our study also revealed a large significant negative relationship between self- esteem, TB, and PB, and a higher correlation coefficient between TB and self-esteem than PB and self-esteem. Given that PB comprises liability and self-hatred as latent constructs, it is surprising that self-esteem is more strongly correlated with TB which comprises loneliness and absence of reciprocal care latent constructs. However, these findings are less surprising given that the same pattern of correlation was found by Marty
(2011). One of her thoughts on this relationship was that the one-item measure of self- esteem used was not sensitive enough to bring about the relationship between PB and self-esteem. By using a validated 10-item self-esteem measure, our study aimed to overcome this limitation. Nevertheless, we also found self-esteem to be more strongly correlated with TB than PB. This suggests that perhaps self-esteem is not the opposite of self-hatred or that self-esteem might also be related to the latent constructs of TB, loneliness and absence of reciprocal care.
Components of self-esteem. The results from our analyses of self-esteem were surprising because contrary to our hypothesis, the usefulness/competence dimension of self-esteem had a smaller relationship with PB compared to the positive self-regard component of self-esteem. The data suggest that individuals are less likely to perceive
62 themselves as a burden when they have a higher level of positive self-regard and that they are more likely to feel as though they belong when they feel useful and/or competence.
We expected that feelings of usefulness/competence would offset PB by tapping into the liability construct of PB. However, it is possible that our results ran contrary to our hypothesis because the positive self-regard dimension of self-esteem loaded more strongly onto the self-hatred construct of PB compared to the usefulness dimension onto the liability construct. Though this is a possible explanation for the results obtained, additional research would be required to confirm how dimensions of self-esteem load onto the different latent constructs of TB and PB.
Neuropsychology and Suicidality
When working with older adults, it is important to conceptualize research from a biopsychosocial framework. From a biological perspective, recent research has begun to elucidate the relationship between cognitive health and suicide among older adults.
Specifically, there is evidence that altered decision-making processes and diminished cognitive control increases an older adult’s vulnerability to suicidal behavior (Kiosses,
Szanto, & Alexopoulos, 2014; Richard-Devantoy, Jollant, Deguigne, & Letourneau,
2013). Recognizing that cognitive impairments may distort social cognition (Szanto et al.,
2012) and psychological insight—potentially having a negative impact on one’s ability to have meaningful social relationships—the IPTS could explain how/why these factors may be linked to SI. Though full neuropsychological testing was beyond the scope of this study, using various cognitive measures from the CATI, the exploratory analyses of this study aimed to investigate the relationship between TB, PB, SI and cognitive factors.
63
The strong positive correlation between SI and the neurocognitive dysfunction scale suggests that as individuals experience greater cognitive impairment their likelihood of experiencing SI increases. Additional analyses revealed that decisional difficulties related to executive functioning were positively correlated with SI. Linear regressions confirmed that in addition to being correlated with SI, both of these variables were predictive of SI.
Though significant correlations for both TB and PB existed between neuropsychological dysfunction, decisional difficulty, and planning, none of these variables were able to independently predict PB, and only the neurocognitive dysfunction scale was a significantly predictive of TB. Given the correlational relationship between TB, PB and cognitive impairment, it is difficult to identify what causes TB and PB in individuals with cognitive decline. Specifically, it is possible that if individuals have impaired social cognition they may misperceive social cues and this may contribute to elevated TB or
PB. Alternatively, it is also possible that individuals are accurately perceiving their social experiences and that their elevated TB and PB are the consequence of a functional decline that is having a negative impact on their social relationships. To explore whether or not elevated TB and PB in the presence of cognitive decline is the result of reduced insight or the social consequence of functional decline, future research may consider including an informant questionnaire that could assess for the presence of TB and PB’s observable indicators.
Data from this study supports Richard-Devantoy et al. (2013) argument that older adults with impaired decision-making processes and diminished cognitive control may increase suicidality, because of the risk of ruminating on social rejection and failure to consider long-term solutions. However, the role of TB and PB within this process is not
64 entirely elucidated though this study. Thus, additional research using measures with greater sensitivity and specificity to certain cognitive domains—in combination with measures of TB and PB—may be able to help clarify this relationship and enhance our understanding of the interaction between neuropsychology and mental health.
Limitations
Though the results of this study have important clinical implications, the limitations of this study must first be addressed. First, this study’s sample lacked ethnic diversity. Specifically, the proportion of White individuals in this sample was notably higher than the national average and slightly higher than the average for Colorado (U.S.
Census Bureau, 2015). As such, the generalizability of these results to the American population may be limited. Furthermore, though White older men that have the highest rates of suicide across the United States, our sample comprised nearly 75% women. The ratio of women to men in this study is notably higher than the distribution in the United
States. Thus, given our high rates of female participants, we cannot confidently state that the results of this study are generalizable to older men in the United States. To address this limitation, future research might consider using purposive sampling rather than a convenience sample to ensure a more representative ratio of men to women.
Another limitation of this study stems from the fact that it was conducted in
Colorado Springs, which is known to have a population that is both highly religious and politically conservative. However, this study failed to include a measure of religiosity and political values. Given that there is an established inverse correlation between religiosity and suicidality, if our sample happened to comprise highly religious individuals, our scores for suicidal ideation may have been lower than those of a less religious sample.
65
This study also demonstrates important limitations because of its structure and the measures used. As a correlational study the results are not causal; consequently, we can only make observations about the relationships between personality, burdensomeness, belongingness, self-esteem, and suicidal ideation rather than being able to state whether or not the variables of interest do in fact cause suicidal ideation. Furthermore, the measures used in this study were all self-report. As such, there is the possibility that the data might be skewed because individuals may have wanted to respond in a socially desirable manner. Self-report measures may also be problematic if individuals lack the cognitive capacity for appropriate psychological insight. As such, there is a risk that individuals might not have answered the questionnaires as accurately as possible.
Additionally, the self-esteem measure used didn’t not demonstrate adequate reliability for the usefulness/competence subscale. As such, we cannot confidently make inferences about the different types of self-esteem and how they express themselves among older adults. Future research may consider doing a more thorough investigation of the reliability of this scale and continue to explore the factors related to self-esteem among older adults. Finally, this study did not include a measure for Acquired Capability, consequently interpretations can only be made about the suicidal ideation component of the IPTS. Given the limited amount of research exploring the construct of AC
(Rimkeviciene, Hawgood, O’Gorman, & Leo, 2015), future studies may consider including Van Orden’s Acquired Capability for Suicide Scale (ACSS; Van Orden, Witte,
Gordon, Bender, & Joiner, 2008) as a way to lend evidence to the ITPS and to advance the clinical utility of the ACSS (Rimkeviciene et al., 2015; Smith, Wolford, Mandracchia,
& Jahn, 2013).
66
Clinical and Practical Implications
Although there are limitations, this study can still contribute to our understanding of suicidal ideation among older adults and more specifically, the relationship between personality traits, and self-esteem with perceptions of burdensomeness and belongingness. This study demonstrated robust relationships between personality and SI.
Since, personality is a relatively stable across the lifespan, it is not an ideal point for intervention. However, clinicians may consider thinking about certain PD traits as risk factors for SI and recognize these traits a cues for SI screening. This study also suggests that separately, self-esteem and TB and PB mediate the relationship between personality and SI. Furthermore, self-esteem is predictive of TB and PB. Therefore, clinicians may consider developing interventions for SI that focus on the different aspects of self-esteem.
By increasing self-esteem, we may be able to decrease levels of TB, and PB, and by reducing TB and PB, this should translate into decreased SI.
Conclusions
Individuals between 50-69 and over the age of 70 experience the highest rates of suicide worldwide (World Health Organization, 2014). Within the next twenty years as the Baby Boomers—a cohort with historically higher rates of suicide—age, the proportion of individuals 65 years and older is expected to approach 20% in the United
States. As such, within the next twenty years we expect to have a higher risk cohort moving into a high risk age bracket. Thus, suicide among older adults is expected to increase and should be considered an important public health problem. The results from this study add to the body of knowledge regarding suicide and older adults. Specifically, within the framework of the IPTS this study elucidates the relationship between TB, PB,
67
SI and normal and abnormal personality traits. Furthermore, this study contributes to a limited body of knowledge regarding self-esteem among older adults and how this relates to TP, PB and SI. It is our hope that by having explored the different factors that relate to and predict TB and PB, the results of this study can be used to improve SI screening and treatment strategies for older adults while also being able to dispel the myth that it normal to wish for death as one ages. REFERENCES
Abrams, R. C., & Horowitz, S. V. (1996). Personality disorders after age 50: A meta- analysis. Journal of Personality Disorders, 10, 271–281. http://doi.org/10.1521/pedi.1996.10.3.271
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Washington, DC: American Psychiatric Association.
Bachman, J. G. (1970). Youth in Transition Volume II the impact of family background and intelligence on tenth-grade boys. Ann Arbor, MI: The Institute for Social Research.
Beck, A. T., Kovacs, M., & Weissman, A. (1979). Assessment of suicidal intention: The Scale for Suicide Ideation. Journal of Consulting and Clinical Psychology, 47, 343–352. http://doi.org/10.1037/0022-006X.47.2.343
Beck, A. T., Schuyler, D., & Herman, I. (1974). Development of suicidal intent scales. In A. T. Beck, H. L. Resnik, & D. J. Lettieri (Eds.), The prediction of suicide (pp. 45–56). Oxford, England: Charles Press Publishers.
Centers for Disease Control and Prevention. (2013). The state of aging and health in America 2013. Atlanta, GA: Centers for Disease Control and Prevention, US Dept of Health and Human Services.
Centers for Disease Control and Prevention. (2014a, December 16). Twenty leading causes of death. Retrieved from http://www.cdc.gov/violenceprevention/suicide/statistics/leading_causes.html
Centers for Disease Control and Prevention. (2014b, December 24). Suicide: Consequences. Retrieved from http://www.cdc.gov/violenceprevention/suicide/consequences.html
Chioqueta, A. P., & Stiles, T. C. (2005). Personality traits and the development of depression, hopelessness, and suicide ideation. Personality and Individual Differences, 38, 1283–1291. http://doi.org/10.1016/j.paid.2004.08.010
Christensen, H., Batterham, P. J., Mackinnon, A. J., Donker, T., & Soubelet, A. (2014). Predictors of the risk factors for suicide identified by the interpersonal- psychological theory of suicidal behaviour. Psychiatry Research, 219, 290–297. http://doi.org/10.1016/j.psychres.2014.05.029 69
Christensen, H., Batterham, P. J., Soubelet, A., & Mackinnon, A. J. (2013). A test of the Interpersonal Theory of Suicide in a large community-based cohort. Journal of Affective Disorders, 144, 225–234. http://doi.org/10.1016/j.jad.2012.07.002
Conwell, Y., Duberstein, P. R., & Caine, E. D. (2002). Risk factors for suicide in later life. Biological Psychiatry, 52, 193–204. http://doi.org/10.1016/S0006- 3223(02)01347-1
Coolidge, F. L., & Merwin, M. M. (1992). Reliability and validity of the Coolidge Axis II Inventory: A new inventory for the assessment of personality disorders. Journal of Personality Assessment, 59, 223–38. http://doi.org/10.1207/s15327752jpa5902_1
Cooper, L. D., Balsis, S., & Oltmanns, T. F. (2014). Aging: Empirical contribution a longitudinal analysis of personality disorder dimensions and personality traits in a community sample of older adults: Perspectives from selves and informants. Journal of Personality Disorders, 28, 151–165. http://doi.org/10.1521/pedi.2014.28.1.151
Cortina, J. M. (1993). What is coefficient alpha? An examination of theory and applications. Journal of Applied Psychology, 78(1), 98–104. http://doi.org/10.1037/0021-9010.78.1.98
Costa, P. T., & McCrae, R. R. (1992). Neo Personality Inventory-Revised (NEO PI-R). Psychological Assessment Resources. Retrieved from http://www.statisticssolutions.com/academic-solutions/resources/directory-of- survey-instruments/neo-personality-inventory-revised-neo-pi-r/
Cramer, R. J., Garza, M. J., Henderson, C. E., Ribeiro, J. D., Silva, C., Smith, A. R., … White, J. (2012). A trait-interpersonal perspective on suicide risk in criminal offenders. Archives of Suicide Research, 16, 334–347. http://doi.org/10.1080/13811118.2013.722057
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297–334. http://doi.org/10.1007/BF02310555
Cukrowicz, K. C., Cheavens, J. S., Van Orden, K. A., Ragain, R. M., & Cook, R. L. (2011). Perceived burdensomeness and suicide ideation in older adults. Psychology and Aging, 26, 331–338. http://doi.org/10.1037/a0021836
Davidson, R., & MacKinnon, J. G. (1996). The size and power of bootstrap test (Working Paper No. 937). Kingston, ON: Department of Economics Queen’s University. Retrieved from https://core.ac.uk/download/files/153/6494167.pdf
Debast, I., Alphen, S. P. J., Van, Rossi, G., Tummers, J. H. A., Bolwerk, N., Derksen, J. J. L., & Rosowsky, E. (2014). Personality traits and personality disorders in late middle and old age: Do they remain stable? A literature review. Clinical Gerontologist, 37, 253–271. http://doi.org/10.1080/07317115.2014.885917 70
De Leo, D., Draper, B. M., Snowdon, J., & Kõlves, K. (2013). Suicides in older adults: A case–control psychological autopsy study in Australia. Journal of Psychiatric Research, 47, 980–988. http://doi.org/10.1016/j.jpsychires.2013.02.009
DeShong, H. L., Tucker, R. P., O’Keefe, V. M., Mullins-Sweatt, S. N., & Wingate, L. R. (2015). Five Factor Model traits as a predictor of suicide ideation and interpersonal suicide risk in a college sample. Psychiatry Research, 226, 217–223. http://doi.org/10.1016/j.psychres.2015.01.002
Draper, B., Kõlves, K., De Leo, D., & Snowdon, J. (2014). A controlled study of suicide in middle-aged and older people: Personality traits, age, and psychiatric disorders. Suicide and Life-Threatening Behavior, 44, 130–138. http://doi.org/10.1111/sltb.12053
Duberstein, P. R., Conwell, Y., & Caine, E. D. (1994). Age differences in the personality characteristics of suicide completers: preliminary findings from a psychological autopsy study. Psychiatry, 57, 213–224.
Duberstein, P. R., Conwell, Y., Seidlitz, L., Denning, D. G., Cox, C., & Caine, E. D. (2000). Personality traits and suicidal behavior and ideation in depressed inpatients 50 years of age and older. Journals of Gerontology Series B, 55(1), P18–P26.
Eades, A. (2014, December). Go on without me: A meta-analytic review of perceived burdensomeness and suicidal ideation in older adults. Unpublished manuscript.
Eades, A., Segal, D. L., Marty, M. A., & Coolidge, F. L. (2015, November). An exploration of personality disorder features and their relationship with burdensomeness and belongingness. Poster presented at the 68th Annual Scientific Meeting of the Gerontological Society of America, Orlando, FL.
Field, A. P. (2013). Discovering statistics using IBM SPSS statistics: And sex and drugs and rock “n” roll. Thousand Oaks, CA: SAGE Publications.
Fiske, A. (2014, September 9). When will we face the facts about suicide in older men? Retrieved from http://psychologybenefits.org/2014/09/09/facts-about-suicide-in- older-men/
Foster, T. (2003). Suicide note themes and suicide prevention. International Journal Of Psychiatry In Medicine, 33, 323–331.
George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and reference, 11.0 update. Boston, MA: Allyn and Bacon.
Harrison, K. E., Dombrovski, A. Y., Morse, J. Q., Houck, P., Schlernitzauer, M., Reynolds, C. F., & Szanto, K. (2010). Alone? Perceived social support and chronic interpersonal difficulties in suicidal elders. International Psychogeriatrics, 22, 445–454. http://doi.org/10.1017/S1041610209991463 71
Harwood, D. M. J., Hawton, K., Hope, T., Harriss, L., & Jacoby, R. (2006). Life problems and physical illness as risk factors for suicide in older people: A descriptive and case-control study. Psychological Medicine, 36, 1265–1274. http://doi.org/10.1017/S0033291706007872
Harwood, D. M. J., Hawton, K., Hope, T., & Jacoby, R. (2001). Psychiatric disorder and personality factors associated with suicide in older people: A descriptive and case- control study. International Journal of Geriatric Psychiatry, 16, 155–165. http://doi.org/http://dx.doi.org/10.1017/S0033291706007872
Hayes, A. (2013). PROCESS [Computer software]. Retrieved from http://processmacro.org/index.html
Heisel, M. J., & Flett, G. L. (2006). The development and initial validation of the Geriatric Suicide Ideation Scale. The American Journal of Geriatric Psychiatry, 14, 742–751. http://doi.org/10.1097/01.JGP.0000218699.27899.f9
Heisel, M. J., Links, P. S., Conn, D., van Reekum, R., & Flett, G. L. (2007). Narcissistic personality and vulnerability to late-life suicidality. The American Journal of Geriatric Psychiatry, 15, 734–741. http://doi.org/10.1097/01.JGP.0000260853.63533.7d
Heisel, M. J., Neufeld, E., & Flett, G. L. (2016). Reasons for living, meaning in life, and suicide ideation: investigating the roles of key positive psychological factors in reducing suicide risk in community-residing older adults. Aging & Mental Health, 20, 195–207. http://doi.org/10.1080/13607863.2015.1078279
Iliceto, P., Fino, E., Sabatello, U., & Candilera, G. (2014). Personality and suicidal ideation in the elderly: Factorial invariance and latent means structures across age. Aging & Mental Health, 18, 792–800. http://doi.org/10.1080/13607863.2014.880404
INSERM Collective Expertise Centre. (2000). Suicide: Psychological autopsy, a research tool for prevention. In INSERM Collective Expert Reports. Paris: Institut national de la santé et de la recherche médicale. Retrieved from http://www.ncbi.nlm.nih.gov/books/NBK7126/
Jahn, D. R., Poindexter, E. K., & Cukrowicz, K. C. (2015). Personality disorder traits, risk factors, and suicide ideation among older adults. International Psychogeriatrics, 27, 1785–1794. http://doi.org/10.1017/S1041610215000174
Joiner, T. E. (2005). Why people die by suicide. Harvard University Press.
Kiosses, D. N., Szanto, K., & Alexopoulos, G. S. (2014). Suicide in older adults: The role of emotions and cognition. Current Psychiatry Reports, 16(11), 1–8. http://doi.org/10.1007/s11920-014-0495-3 72
Kjølseth, I., Ekeberg, Ø., & Steihaug, S. (2009). “Why do they become vulnerable when faced with the challenges of old age?” Elderly people who committed suicide, described by those who knew them. International Psychogeriatrics, 21, 903–912. http://doi.org/10.1017/S1041610209990342
Lester, D., & Gunn, J. F. (2012). Perceived burdensomeness and thwarted belonging: An investigation of the interpersonal theory of suicide. Clinical Neuropsychiatry, 9, 221–224.
Marty, M. A. (2011). Older adults and the desire for suicide: The role of perceived burdensomeness and thwarted belongingness (Unpublished doctoral dissertation). University of Colorado at Colorado Springs, Colorado Springs.
Marty, M. A., Segal, D. L., Coolidge, F. L., & Klebe, K. J. (2012). Analysis of the psychometric properties of the Interpersonal Needs Questionnaire (INQ) among community-dwelling older adults. Journal of Clinical Psychology, 68, 1008– 1018. http://doi.org/10.1002/jclp.21877
Mordekar, A., & Spence, S. A. (2008). Personality disorder in older people: How common is it and what can be done? Advances in Psychiatric Treatment, 14, 71– 77. http://doi.org/10.1192/apt.bp.107.003897
Newton-Howes, G., Clark, L. A., & Chanen, A. (2015). Personality disorder across the life course. The Lancet, 385(9969), 727–734. http://doi.org/10.1016/S0140- 6736(14)61283-6
O’Connor, R. C., & Nock, M. K. (2014). The psychology of suicidal behaviour. The Lancet Psychiatry, 1(1), 73–85. http://doi.org/10.1016/S2215-0366(14)70222-6
Phillips, J. A., Robin, A. V., Nugent, C. N., & Idler, E. L. (2010). Understanding recent changes in suicide rates among the middle-aged: Period or cohort effects? Public Health Reports, 125, 680.
Ranzijn, R., Keeves, J., Luszcz, M., & Feather, N. T. (1998). The role of self-perceived usefulness and competence in the self-esteem of elderly adults: Confirmatory factor analyses of the Bachman Revision of Rosenberg’s Self-Esteem Scale. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, 53B(2), P96–P104. http://doi.org/10.1093/geronb/53B.2.P96
Reynolds, K., Pietrzak, R. H., El-Gabalawy, R., Mackenzie, C. S., & Sareen, J. (2015). Prevalence of psychiatric disorders in U.S. older adults: Findings from a nationally representative survey. World Psychiatry, 14(1), 74–81. http://doi.org/10.1002/wps.20193
73
Richard-Devantoy, S., Jollant, F., Deguigne, F., & Letourneau, G. (2013). Marqueurs neurocognitifs de la vulnérabilité suicidaire de la personne âgée: Revue de la littérature. [Neurocognitive markers of suicide vulnerability in the elderly: A review]. Gériatrie et Psychologie Neuropsychiatrie Du Vieillissement, 11, 367– 378.
Rimkeviciene, J., Hawgood, J., O’Gorman, J., & Leo, D. D. (2015). Assessment of acquired capability for suicide in clinical practice. Psychology, Health & Medicine, 1–10. http://doi.org/10.1080/13548506.2015.1115108
Robins, R. W., Tracy, J. L., Trzesniewski, K., Potter, J., & Gosling, S. D. (2001). Personality correlates of self-esteem. Journal of Research in Personality, 29, 463– 482.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton University Press Princeton, NJ. Retrieved from http://psycnet.apa.org/journals/ort/36/3/560.pdf%26productCode=pa
Saucier, G. (1994). Mini-markers: A brief version of Goldberg’s unipolar big-five markers. Journal of Personality Assessment, 63, 506–516. http://doi.org/10.1207/s15327752jpa6303_8
Segal, D. L., Coolidge, F. L., & Rosowsky, E. (2006). Personality disorders and older adults: Diagnosis, assessment, and treatment. John Wiley & Sons: Hoboken, NJ.
Segal, D. L., Gottschling, J., Marty, M., Meyer, W. J., & Coolidge, F. L. (2015). Relationships among depressive, passive-aggressive, sadistic and self-defeating personality disorder features with suicidal ideation and reasons for living among older adults. Aging & Mental Health, 19, 1071–1077. http://doi.org/10.1080/13607863.2014.1003280
Segal, D. L., Marty, M. A., Meyer, W. J., & Coolidge, F. L. (2012). Personality, suicidal ideation, and reasons for living among older adults. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 67, 159–166. http://doi.org/10.1093/geronb/gbr080
Silva, C., Ribeiro, J. D., & Joiner, T. E. (2015). Mental disorders and thwarted belongingness, perceived burdensomeness, and acquired capability for suicide. Psychiatry Research, 226, 316–327. http://doi.org/10.1016/j.psychres.2015.01.008
Smith, P. N., Wolford, C., Mandracchia, J. T., & Jahn, D. R. (2013). An exploratory factor analysis of the Acquired Capability for Suicide Scale in male prison inmates. Psychological Services, 10, 97–105. http://doi.org/10.1037/a0030817
Statistics Canada. (2010). Population projections for Canada, provinces and territories: Highlights. Retrieved July 24, 2014, from http://www.statcan.gc.ca/pub/91-520- x/2010001/aftertoc-aprestdm1-eng.htm 74
Szanto, K., Dombrovski, A. Y., Sahakian, B. J., Mulsant, B. H., Houck, P. R., Reynolds III, C. F., & Clark, L. (2012). Social emotion recognition, social functioning, and attempted suicide in late-life depression. The American Journal of Geriatric Psychiatry, 20, 257–265. http://doi.org/10.1097/JGP.0b013e31820eea0c
Szanto, K., Reynolds III, C. F., Frank, E., Stack, J., Fasiczka, A. L., Miller, M., … Kupfer, D. J. (1996). Suicide in elderly depressed patients: Is active vs. passive suicidal ideation a clinically valid distinction? The American Journal of Geriatric Psychiatry, 4, 197–207. http://doi.org/10.1097/00019442-199622430-00003
Turvey, C. L., Conwell, Y., Jones, M. P., Phillips, C., Simonsick, E., Pearson, J. L., & Wallace, R. (2002). Risk factors for late-life suicide: A prospective, community- based study. The American Journal of Geriatric Psychiatry, 10, 398–406. http://doi.org/10.1097/00019442-200207000-00006
U.S. Census Bureau. (2015). Population estimates, July 1, 2015, (V2015). Retrieved from //www.census.gov/quickfacts/
Useda, J. D., Duberstein, P. R., Conner, K. R., & Conwell, Y. (2004). Personality and attempted suicide in depressed adults 50 years of age and older: A facet level analysis. Comprehensive Psychiatry, 45, 353–361. http://doi.org/10.1016/j.comppsych.2004.06.002
Van Orden, K. A. (2009). Construct validity of the interpersonal needs questionnaire (Doctoral dissertation). Florida State University. Retrieved from http://diginole.lib.fsu.edu/islandora/object/fsu:182668/datastream/PDF/download/ citation.pdf
Van Orden, K. A., Cukrowicz, K. C., Witte, T. K., & Joiner, T. E. (2012). Thwarted belongingness and perceived burdensomeness: Construct validity and psychometric properties of the Interpersonal Needs Questionnaire. Psychological Assessment, 24, 197–215. http://doi.org/10.1037/a0025358
Van Orden, K. A., O’Riley, A. A., Simning, A., Podgorski, C., Richardson, T. M., & Conwell, Y. (2014). Passive suicide ideation: An indicator of risk among older adults seeking aging services? The Gerontologist, 5, 972-980. http://doi.org/10.1093/geront/gnu026
Van Orden, K. A., Smith, P. N., Chen, T., & Conwell, Y. (2016). A case controlled examination of the Interpersonal Theory of Suicide in the second half of life. Archives of Suicide Research, 20, 323–335. http://doi.org/10.1080/13811118.2015.1025121
Van Orden, K. A., Wiktorsson, S., Duberstein, P., Berg, A. I., Fässberg, M. M., & Waern, M. (2015). Reasons for attempted suicide in later life. The American Journal of Geriatric Psychiatry, 23, 536–544. http://doi.org/10.1016/j.jagp.2014.07.003 75
Van Orden, K. A., Witte, T. K., Cukrowicz, K. C., Braithwaite, S. R., Selby, E. A., & Joiner, T. E. (2010). The interpersonal theory of suicide. Psychological Review, 117, 575–600. http://doi.org/10.1037/a0018697
Van Orden, K. A., Witte, T. K., Gordon, K. H., Bender, T. W., & Joiner, T. E. (2008). Suicidal desire and the capability for suicide: Tests of the Interpersonal- Psychological Theory of Suicidal behavior among adults. Journal of Consulting and Clinical Psychology, 76, 72-83.
Vanyukov, P. M., Szanto, K., Hallquist, M., Moitra, M., & Dombrovski, A. Y. (2016). Perceived burdensomeness is associated with low-lethality suicide attempts, dysfunctional interpersonal style, and younger rather than older age. International Journal of Geriatric Psychiatry. http://doi.org/10.1002/gps.4526
World Health Organization. (2014). Preventing suicide: A global imperative. Retrieved from http://www.who.int/mental_health/suicide- prevention/world_report_2014/en/
Yesavage, J. A., Brink, T. L., Rose, T. L., Lum, O., Huang, V., Adey, M., & Leirer, V. O. (1982). Development and validation of a geriatric depression screening scale: A preliminary report. Journal of Psychiatric Research, 17(1), 37–49.
Yesavage, J. A., & Sheikh, J. I. (1986). 9/Geriatric Depression Scale (GDS) recent evidence and development of a shorter violence. Clinical Gerontologist, 5, 165– 173.
76
APPENDIX
Institutional Review Board Approval Notice
Institutional Review Board (IRB) for the Protection of Human Subjects Date: 1/19/2016 IRB Review
IRB PROTOCOL NO.: 16-110 Protocol Title: Personality Traits and Attitudes of Older Adults Toward Relationships, Life, and Death Principal Investigator: Allison Eades Faculty Advisor if Applicable: Daniel Segal Application: New Application Type of Review: Expedited 7 Risk Level: No more than Minimal Risk Renewal Review Level (If changed from original approval) if Applicable: N/A No Change This Protocol involves a Vulnerable Population: N/A (No Vulnerable Population) Expires: 18 January 2017 *Note, if exempt: If there are no major changes in the research, protocol does not require review on a continuing basis by the IRB. In addition, the protocol may match more than one review category not listed. Externally funded: ☒ No ☐ Yes OSP #: Sponsor:
Thank you for submitting your Request for IRB Review. The protocol identified above has been reviewed according to the policies of this institution and the provisions of applicable federal regulations. The review category is noted above, along with the expiration date, if applicable.
Once human participant research has been approved, it is the Principal Investigator’s (PI) responsibility to report any changes in research activity related to the project: x The PI must provide the IRB with all protocol and consent form amendments and revisions. x The IRB must approve these changes prior to implementation. x All advertisements recruiting study subjects must also receive prior approval by the IRB. x The PI must promptly inform the IRB of all unanticipated serious adverse (within 24 hours). All unanticipated adverse events must be reported to the IRB within 1 week (see 45CFR46.103(b)(5). Failure to comply with these federally mandated responsibilities may result in suspension or termination of the project. x Renew study with the IRB prior to expiration. x Notify the IRB when the study is complete
If you have any questions, please contact Research Compliance Specialist in the Office of Sponsored Programs at 719-255-3903 or [email protected]
Thank you for your concern about human subject protection issues, and good luck with your research.
Sincerely yours, Zek Cypress Valkyrie
Zek Cypress Valkyrie, PhD IRB Reviewer www.uccs.edu/~osp/compliance/ 1420 Austin Bluffs ParkwayColorado Springs, CO 80918 719-255-3321 phone 719-255-3706 fax Version 2/12/13