Perceived Burdensomeness as an Indicator of Suicidal Symptoms
- First Online:
- Cite this article as:
- Van Orden, K.A., Lynam, M.E., Hollar, D. et al. Cogn Ther Res (2006) 30: 457. doi:10.1007/s10608-006-9057-2
- 497 Views
The Interpersonal-Psychological Theory of Suicidal Behavior proposes precursors to serious suicidality, including the perception that one is a burden on loved ones. The purpose of the present study was to evaluate the association of perceived burdensomeness and key suicide-related variables in 343 adult outpatients of the Florida State University Psychology Clinic (187 female; 156 male). Participants completed the Beck Scale for Suicide Ideation and the Beck Depression Inventory, as well as items on perceived burdensomeness and hopelessness. Perceived burdensomeness remained a significant predictor of suicidality indicators (i.e., attempt status and BSSI scores) above and beyond the contribution one of the most robust predictors of suicidality, hopelessness. Results suggest that both burdensomeness and hopelessness display predictive power with regards to suicidal behavior and that perceived burdensomeness displayed the signature of a resilient suicide risk factor. Thus, targeting perceived burdensomeness in the assessment of suicidal behavior may aid in treatment and prevention efforts.
The Interpersonal-Psychological Theory of Suicidal Behavior (Joiner, 2005) proposes three precursors to serious suicidality: (1) the acquired capability to self-injure; (2) thwarted belongingness; and (3) the perception that one is a burden on loved ones. This theory provides the framework on which this paper will center, with specific attention to the last aspect of the theory—perceived burdensomeness. The theory posits that perceived burdensomeness involves perceptions that the self is so incompetent as to be a liability for others.
Various lines of evidence implicate perceived burdensomeness in suicidal behavior. Suicide among the Yuit Eskimos of St. Lawrence Island provides a persuasive anecdotal example. Among this group, to become too sick, infirm, or old may threaten the group’s survival (i.e., burden the group), the explicit and socially sanctioned response to which was ritual suicide. The ritual was graphic, often involving the family members’ participation in the shooting or hanging of the victim (Maris, Berman, & Silverman, 2000; this pattern applied to other Eskimo groups as well; Leighton & Hughes, 1955).
To our knowledge, four studies have been framed as direct empirical tests of the role of perceived burdensomeness in suicidality. In two studies, Joiner et al. (2002) trained raters to evaluate suicide notes regarding the following dimensions: perceived burdensomeness, hopelessness, and generalized emotional pain. In Study 1, unknown to the raters, half of the notes were from people who died by suicide, and half were from people who attempted suicide and survived. Notes from those who died by suicide contained more perceived burdensomeness than notes from attempters; no effects were found regarding hopelessness and emotional pain. A second study in the Joiner et al. report took a similar approach, except that all notes were from those who died by suicide, and perceived burdensomeness, hopelessness, and generalized emotional pain were used as predictors of lethality of suicide method. Here again, perceived burdensomeness was a significant predictor of lethality, whereas hopelessness and generalized emotional pain were not. Important features of this report included Study 1’s relatively stringent comparison of notes from those who died by suicide to notes from those who attempted and survived (this distinction was unknown to raters), and the simultaneous analysis of perceived burdensomeness, hopelessness, and generalized emotional pain, with perceived burdensomeness emerging as the only unique predictor of completer versus attempter status (Study 1) and lethality of method (Study 2).
DeCatanzaro (1995) conducted a survey on suicidal ideation and quality of family contacts among several hundred community participants, as well as on five high-suicide-risk groups (e.g., general psychiatric patients, incarcerated psychiatric patients). Within each of these samples, of all the many variables assessed, perceived burdensomeness toward family was especially correlated with suicidal ideation. Finally, Brown, Dahlen, Mills, Rick, and Biblarz, (1999) report that feelings of being a burden on kin stood out as a unique and specific predictor of suicide-related symptoms even when other key variables were controlled.
Though the studies reviewed above are consistent with a role for perceived burdensomeness in suicidality, relatively little work has been conducted on the topic. Moreover, of the work that has been done to date, relatively little has examined clinical samples. Accordingly, the purpose of the present study was to evaluate the association of perceived burdensomeness and key suicide-related variables in a relatively large clinical sample of adult psychotherapy patients. It was hypothesized that perceived burdensomeness would directly relate to past number of suicide attempts as well as to an index of current suicidal symptoms, and furthermore, that this relationship would exist even when controlling for known risk factors such as personality disorder status, depressive symptoms, and hopelessness. The relationship between perceived burdensomeness and suicidality indicators above and beyond the contribution of hopelessness will be examined. If perceived burdensomeness is a predictor of suicidality indicators, its associations to suicidality should be similarly rigorous as those regarding the documented risk factor of hopelessness (cf. Brown, Beck, Steer, & Grisham, 2000) and it should predict suicidality indicators when controlling for levels of hopelessness.
Participants and setting
Participants were 343 (187 female; 156 male) adult outpatients of the Florida State University Psychology Clinic ranging in age from 18 to 62; mean age for the sample was 26.50 (SD = 9.38). The ethnic composition of this sample, which was roughly representative of the community of Tallahassee, Florida, was 83% Caucasian, 8% African–American, 4.7% Hispanic, 2.3% Asian–American, 0.3% Native American; 2.4% declined to indicate race. Areas of diagnosis were represented in the following proportions: 39% mood disorder, 14.6% anxiety disorder, 6% substance use disorder, 12.2% personality disorder, 9% adjustment disorder, and 18% other disorders.
The FSU Psychology Clinic, although university affiliated, primarily serves patients not affiliated with FSU who present with clinical disorders typical of a community mental health outpatient clinic. The FSU Psychology Clinic employs minimal exclusionary criteria, referring elsewhere only people with psychotic or bipolar-spectrum disorders who are not stabilized on medications and those who are a clear and imminent physical danger to self or others. Moreover, because the FSU Psychology Clinic has an inexpensive sliding fee scale, a relatively low SES sample is served.
Participants at intake were given the Beck Scale for Suicide Ideation (BSSI) and the Beck Depression Inventory (BDI). Important for this study, an item on perceived burdensomeness and an item on hopelessness were also administered. The participants completed standard background information forms, including a question on past suicide attempts. Consistent with Rudd, Joiner, and Rajab (1996), this variable was coded such that 0 = no past attempts, 1 = one past attempt, 2 = 2 or more past attempts. About 281 patients had never attempted suicide; 42 had attempted once; and 20 had attempted twice or more.
All participants signed written consent forms acknowledging that their responses will be used for the purposes of research in accordance with standards set by the FSU IRB. Rating scales were administered by clinic staff, who are thoroughly trained and closely monitored and supervised.
I do not think that my relatives would be happier if I were gone.
I think that my relatives would be happier if I were gone.
I am sure that my relatives would be happier if I were gone.
There is something I can do to make things better for myself.
I am unsure that there is something I can do to make things better for myself.
I might as well give up because there is nothing I can do about making things better for myself.
Beck Scale for Suicide Ideation (BSSI; Beck & Steer, 1991). The BSSI is a 21 item self report inventory designed for the assessment of suicide intention. A score of 0–2 is assigned for each item and total scores for the BSSI range from 0 to 42 where an increase in score represents a higher level of suicidal ideation and possible intent.
Beck Depression Inventory (BDI; Beck, Rush, Shaw & Emery, 1979). Depression levels of participants were assessed using the BDI. The BDI is a frequently used 21 item self-report measure of depressive symptoms where scores for each item range from 0 to 3. Total scores for the BDI range from 0 to 63 where higher scores represent a higher occurrence of symptoms associated with depression.
Personality Disorder Status. Therapists working under the close supervision of clinical psychologists diagnosed the participants in the study. Diagnoses at the FSU clinic are made using a consensus approach (through group supervision) that utilizes all available information at intake. These chart diagnoses of Axis II disorders were established using clinical interviewing procedures highly similar to the Structured Clinical Interview for the DSM-II (SCID-II; First, Spitzer, Gibbon, & Williams, 1995). The data for the psychometric properties of the SCID are abundant, and there are various data to support the concordance between chart diagnoses and diagnoses based on structured clinical interview. For example, Fennig, Craig, Tanenberg-Karant, and Bromet (1994) reported the following information regarding the relationship between SCID and chart depressive diagnoses in academic medical centers: Sensitivity = .73, Specificity = .95, Kappa = .64. In addition, Lonigan, Carey, and Finch (1994) reported a relatively high agreement between the structured research diagnoses and chart diagnoses of depression or anxiety in a sample of inpatient children. Other studies support the construct validity of these diagnoses (Joiner, Katz, & Lew, 1997; Joiner & Lonigan, 2000). This variable was coded such that 0 = personality disorder absent, 1 = personality disorder present. About 12.2% of the sample was diagnosed with a personality disorder.
We were interested in the association between perceived burdensomeness and two indices of suicidality, past suicide attempts and current suicidal symptoms. We sought to evaluate the strength of this association when the following variables were controlled: demographic variables of age and gender, depressive symptoms, and personality disorder status, as well as a strong predictor of suicidality, hopelessness. To evaluate the relation between perceived burdensomeness, past suicide attempts, and current suicidal symptoms, two regression equations were constructed with past suicide attempts as one dependent variable and current suicidal symptoms as the other dependent variable. To statistically control for age, gender, and personality disorder status, these variables were entered into the equation in the first step. Depression scores were entered in the second step and hopelessness scores in the third step. Finally, to evaluate if perceived burdensomeness predicts suicidality indicators above and beyond hopelessness (as well as the other variables included in the model), perceived burdensomeness was entered in the last step. If perceived burdensomeness does contribute to the prediction of suicidality indicators above and beyond the contribution of hopelessness, the amount of unique variance on the suicidality indicators accounted for by perceived burdensomeness (i.e., the semi-partial r2) should emerge as significant in the models.
Means and standard deviations for, and intercorrelations between, all measures
3. Suicide Attempt Status
4. BSSI Score
5. BDI Score
6. PD Status
Perceived burdensomeness, hopelessness, suicide attempt status, and BSSI and BDI scores were all significantly intercorrelated. It is of note that perceived burdensomeness was correlated with age, such that older people felt more of a burden. Of course, for the present purpose, the associations between perceived burdensomeness on the one hand and suicide attempt status (r = .21, P < .05) and BSSI suicidality scores (r = .32, P < .05) on the other hand were of particular interest, as were the correlations between hopelessness on the one hand and suicide attempt status (r = .20, P < .05) and BSSI suicidality scores (r = .46, P < .05) on the other hand.
Does perceived burdensomeness predict suicide attempt status above and beyond the contribution of depressive symptoms and hopelessness?
Hierarchical multiple regression equation predicting current suicidal symptoms
Predictors entered in set
F for set
R2 for set
t for Predictors in Set
Semi-partial correlation (sr)
In the final (and critical) step, a model containing perceived burdensomeness as well as hopelessness, depressive symptoms, and the covariates significantly predicted suicidal symptoms (F(6, 336) = 39.37, P < .001) and accounted for 41% of the variance in suicidal symptoms. The addition of perceived burdensomeness to the model accounted for an additional 3% of the variance in suicidal symptoms (above and beyond the contribution of depressive symptoms and hopelessness), with an associated f2 effect size of .05—a small to medium effect. Higher levels of perceived burdensomeness significantly predicted higher levels of suicidal symptoms (sr = .17, t = 3.94, P < .001). This result suggests that perceived burdensomeness was predictive of the level of current suicidal symptoms above and beyond the contribution of depressive symptoms and hopelessness, two relatively powerful suicide-related variables.
Does perceived burdensomeness predict current suicidality symptoms above and beyond the contribution of depressive symptoms and hopelessness?
Hierarchical multiple regression equation predicting suicide attempt status
Predictors entered in set
F for set
t for Predictors in set
Semi-partial correlation (sr)
In the final (and critical) step, a model containing perceived burdensomeness as well as hopelessness, depressive symptoms, and the covariates significantly predicted attempt status (F(6, 336) = 6.02, P < .001) and accounted for 10% of the variance in attempt status. The addition of perceived burdensomeness to the model accounted for an additional 2% of the variance in attempt status (above and beyond the contribution of depressive symptoms and hopelessness), with an associated f2 effect size of .02—a small effect. Higher levels of perceived burdensomeness significantly predicted greater numbers of past attempts (sr = .14, t = 2.74, P < .01). This result suggests that perceived burdensomeness was predictive of suicide attempt status above and beyond the contribution of depressive symptoms and hopelessness, two relatively powerful suicide-related variables.1
Perceived burdensomeness remained a significant predictor of suicidality indicators (i.e., attempt status and BSSI scores) above and beyond the contribution one of the most robust predictors of suicidality, hopelessness. Although he predictive power of perceived burdensomeness appears to be stronger for current suicidal symptoms than for number of past attempts (i.e., the percentage of variance accounted for by the final model containing perceived burdensomeness was 41% for current symptoms and 10% for past attempt status), perceived burdensomeness did account for unique variance on both current and past suicidality. These analyses suggest that perceived burdensomeness may explain unshared variance in suicidality symptoms not better accounted for by depressive symptoms and hopelessness. Both hopelessness and perceived burdensomeness predicted suicidal behavior above and beyond the contribution of depressive symptoms. Thus, results suggest that both burdensomeness and hopelessness display predictive power with regards to suicidal behavior and that perceived burdensomeness displayed the signature of a resilient suicide risk factor.
It is important to note potential psychometric limitations of some of the measures used in the current study. We do not have data in direct support of the reliability and validity of the Axis II diagnoses used in the current study: while research supports the concordance between chart diagnoses and diagnoses made using structured clinical interviews for Axis I diagnoses (e.g., Fennig, Craig, Tanenberg-Karant, & Bromet, 1994), research is needed to directly support the concordance between chart diagnoses and diagnoses made using structured clinical interviews for Axis II diagnoses. An additional limitation is the use of one-item measures of perceived burdensomeness and hopelessness, both of which are likely to be less reliable estimates of their respective constructs than multi-item measures.
Although future research directly addressing this issue is needed, we suggest that the convergence of these results with past research provides some support for the construct validity of the one-item measure of perceived burdensomeness. For example, in the Joiner et al. (2002) studies, suicide notes were rated for extent of perceived burdensomeness with the following question: “To what degree does the passage imply the idea that ‘my loved ones will be better off when I’m gone?’” This item was rated with high inter-rater reliability. Moreover, in the first study from that report (with notes from individuals who died by suicide and notes from individuals who attempted suicide), only levels of perceived burdensomeness (not hopelessness or emotional pain) were rated to be higher in notes of individuals who died by suicide (compared to attempters who survived). In the second study, only perceived burdensomeness significantly predicted lethality of method (not hopelessness or emotional pain). The Joiner et al. (2002) studies used a different method to assess perceived burdensomeness than the one used in the current study (i.e., ratings versus self-report), but used a similar stimulus item to assess the construct (i.e., perceptions that loved ones would be better off). The convergence of results from the Joiner et al. studies and the present study (i.e., the ability of perceived burdensomeness to predict suicide attempt status even when hopelessness is covaried) using different methods provides support for the validity of the burdensomeness measure used in the present study. In addition, both constructs in the present study (i.e., burdensomeness and hopelessness) were assessed with single-item indices to place them on similar footing.
Future research is needed to support the construct validity of the one-item measure of hopelessness. Our measure was designed to reflect pessimistic beliefs about one’s future. However, because the measure involved beliefs about self-efficacy (i.e., “something I can do to make things better for myself”) it might reflect perceptions of problem-solving abilities. Given that poorer self-appraised problem solving abilities have been found to predict higher levels of suicidal symptoms (Wingate, Van Orden, Joiner, Williams, & Rudd,2005), it is unclear from our findings whether perceived burdensomeness predicts suicidality indices above and beyond the contribution of hopelessness about the future in general or only hopelessness about one’s ability to improve the future. Due to the strong association between self-appraised problem-solving abilities and suicidal symptoms (Wingate et al., found a correlation magnitude of .49), we suggest that the inclusion of a measure of hopelessness involving self-efficacy provides an especially stringent test of our hypothesis: that perceived burdensomeness predicted suicidality indicators above and beyond the contribution of a form of hopelessness found to strongly predict suicidal symptoms suggests strong predictive utility for perceived burdensomeness in the domain of suicidality.
An additional limitation of the current study is its cross-sectional design which precludes conclusions about causal relations between perceived burdensomeness and suicidality. It may very well be the case that persons who have multiple suicide attempts (or high levels of suicidal symptoms) may come to view themselves as a burden to others. Thus, future research utilizing a longitudinal design is needed to further clarify the nature of the relationship between perceived burdensomeness and the development of suicidal behaviors.
It is important to use caution in generalizing the results from this study because the sample was predominantly Caucasian. This limits generalizability essentially to European American populations, and extension of this work to samples with significant representation from other ethnic minority groups is encouraged. Additionally, future research is needed to clarify the nature of the relationship between perceived burdensomeness and hopelessness in the prediction of suicidality: future studies could address whether the combination of burdensomeness and hopelessness is an especially pernicious combination for the development of suicidal behavior.2
The results of the current study, taken together with those obtained by others (e.g., DeCatanzaro, 1995; Joiner et al., 2002), suggest that clinical assessment of perceived burdensomeness as a predictor of increased risk for suicidal behavior may be indicated. Our findings also suggest that perceived burdensomeness may be a useful focus for crisis intervention. Rudd, Joiner, and Rajab (2001) described techniques for suicide crisis intervention, including a “symptom matching hierarchy” which is created by having the client list his/her most painful symptoms, feelings, or attitudes—perceived burdensomeness (if present) should be targeted in this list. The therapist and client then come up with concrete recommendations for each of the top two or three symptoms in order to take the edge off the problem so that a client is a little more comfortable and thus better able to work toward solving the underlying problem. To ameliorate perceptions of burdensomeness the client and therapist might create a list of ways the client has contributed to the lives of other people as well as society, both in the present and in the past. The client may then be given the list to take home and be instructed to elaborate and expand the list, perhaps by writing a short paragraph explaining his/her contributions to others and corresponding feelings of effectiveness.
Perceptions of burdensomeness are also potential targets for intervention within Rudd et al.’s (2001) cognitive therapy for suicidal behavior. Cognitive restructuring of negative thoughts is a focus of the intervention; our results suggest that thoughts related to perceived burdensomeness should be prioritized. An identified thought, such as “I am a burden on the people in my life” can be connected to the type of cognitive distortion it represents; in this case, the thought involves “labeling”—putting a stable, global label on oneself without enough data to back up such a label. Client and therapist would then collect data related to the thought (e.g., the list of ways the client has contributed to others, described above). Next, these data would then be used to restructure the thought to make it less global and less stable—and more in line with the collected data (e.g., “Even though I feel like a burden on others at times, I actually contribute to the well-being of others in multiple ways”). Finally, the client then behaves as if the restructured thought is true: perhaps by seeking out concrete ways to contribute to the lives of others while also acknowledging that most people will occasionally feel like a burden. Our results indicating that perceived burdensomeness was a significant predictor of suicidality indicators suggests that future research on the role of perceived burdensomeness in the development and maintenance of suicidal symptoms is a promising avenue for the areas of suicide risk assessment, crisis intervention, and treatment.
Two models which included a quadratic term for burdensomeness (i.e., burdensomeness-squared) were run to examine of the form of the relationship between burdensomeness and (1) current suicidal symptoms and (2) past suicide attempts. In the final step of both models, burdensomeness and burdensomeness-squared remained significant predictors of current suicidality, suggesting the potential for a curvilinear relationship. However, the burdensomeness main effect (our effect of interest) remained significant in both models after the inclusion of burdensomeness-squared. Future studies could further clarify the form of the relationship between perceived burdensomeness and suicidal symptoms.
Two additional models were run to examine the relationship between perceived burdensomeness, hopelessness, and suicidality. The first model included the interaction of perceived burdensomeness and hopelessness in the final step in the prediction of current suicidal symptoms; this interaction was significant, suggesting the need for additional studies to examine the interactive effects of perceived burdensomeness and hopelessness in the prediction of current suicidal symptoms. The second model included the interaction of perceived burdensomeness and hopelessness in the final step in the prediction of past attempts; this interaction was not significant. Additional studies are needed to replicate this effect as well as investigate potential explanations for the lack of interaction between perceived burdensomeness and hopelessness for past suicide attempts and the presence of an interaction for current suicidal symptoms.