Perceived burdensomeness, thwarted belongingness, and suicide ideation: Re-examination of the Interpersonal-Psychological Theory in two samples
Introduction
Suicides in the Western Hemisphere have increased markedly over the past decade, with over 40,000 deaths in the United States in 2012 alone (Drapeau and McIntosh, 2014, Reeves et al., 2012). Even more prevalent than lethal attempts is their precursor: suicide ideation. Predictive of suicide attempts, even in the absence of planning, suicide ideation is estimated to affect 2–10% of United States adults annually and thus represents a primary target for treatment and prevention efforts (Nock et al., 2008).
At the forefront of such efforts is the Interpersonal-Psychological Theory of Suicide (IPTS; Joiner, 2005, Van Orden et al., 2010), which distinguishes between two aspects of suicide risk. First, due to the fact that lethal suicide attempts involve methods that are both fear-inducing and painful, the IPTS argues that an acquired capability for suicide must be developed prior to such an attempt. This capability, comprised of increased fearlessness about death and pain tolerance, is thought to develop from habituation to painful and provocative events.
The second aspect of suicide risk and the focus of this investigation is suicide ideation, which comes in both passive (i.e., thinking one would be better off dead) and active (i.e., desire to bring about one׳s death) forms. According to the IPTS, suicide ideation is driven by perceived burdensomeness (the perception that one is a liability on others), thwarted belongingness (lack of reciprocal caring relationships), and hopelessness about those states. Specifically, the presence of either perceived burdensomeness or thwarted belongingness is believed to predict passive ideation, whereas the mutual presence of these risk factors is believed to predict active ideation. In this way, perceived burdensomeness and thwarted belongingness are believed to be (a) sufficient proximal causes of suicide ideation and (b) synergistic in their effect, with increases in one variable amplifying the effect of the other, an assumption of the IPTS we hence refer to as the synergy hypothesis.
Evaluation of the synergy hypothesis is ongoing, with many studies finding support for an interaction between perceived burdensomeness and thwarted belongingness across diverse samples and research contexts. These include samples of young, middle-aged, and older adults (Christensen et al., 2013, Cukrowicz et al., 2013, Joiner et al., 2009, Van Orden et al., 2008), diverse ethnic samples (Davidson et al., 2010, O׳Keefe et al., 2014, Wong et al., 2011), and psychiatric inpatients (Monteith et al., 2013). Conversely, several studies have failed to detect an interaction. For example, Garza and Pettit (2010) did not find a significant interaction between perceived burdensomeness and familism, a culturally-relevant thwarted belongingness construct in Mexican and Mexican–American women. Moreover, Bryan et al. (2010) did not find support for the synergy hypothesis in a sample of active duty Airmen, nor did they find lower-order effects of either perceived burdensomeness or thwarted belongingness. Elsewhere, Hill et al. (2015) failed to detect a significant interaction between perceived burdensomeness and thwarted belongingness in a sample of psychiatric inpatients. Thus, the overall support for the synergy hypothesis is best characterized as modest, and an explanation for variability in that support is becoming increasingly important.
One potential explanation for this inconsistency is that previous models have not controlled for quadratic effects, which can distort the sign, size, and significance of interaction terms when mistakenly neglected (Cortina, 1993, Ganzach, 1997, Ganzach, 1998). That is, while some previous research has considered nonlinear link functions relating ideation and its predictors (e.g., logistic regression by Joiner et al. (2009); zero-inflated negative-binomial regression by Cukrowicz et al. (2013)), to our awareness, no previous study has considered the specific possibility that perceived burdensomeness and thwarted belongingness may be quadratically related to ideation, in addition to interactive. In this case, that possibility is uniquely important to examine due to the high correlation that is often observed between perceived burdensomeness and thwarted belongingness (e.g., r>0.70 Anestis and Joiner, 2011, Tucker et al., 2013).
To explain, the likelihood of an illusory interaction becoming erroneously significant increases as the two constituent predictors become more correlated (Ganzach, 1997, Ganzach, 1998). This is because as the two constituent predictors become more correlated, their respective quadratic terms also become increasingly similar to their interaction term. Thus, if a true quadratic term is left out of a model, its explanatory power (i.e., the variance it shares with the outcome variable) can be erroneously attributed to an otherwise irrelevant interaction (Cortina, 1993).
Notably, neglecting quadratic terms can also impair the power to detect an interaction. In fact, simulation research has shown that mistaken exclusion of quadratic terms can yield a Type II error rate of 96% for true interaction terms under some conditions (Ganzach, 1997, Ganzach, 1998). Likewise, neglect of quadratic effects can flip the sign of a significant interaction, causing a synergistic relationship to look like a dampening one, or the reverse (Ganzach, 1997).
As mentioned above, the synergy hypothesis is at unique risk for the confounding effect of neglected quadratic terms due to the fact that perceived burdensomeness and thwarted belongingness are often highly correlated (Cortina, 1993). This makes a formal test of quadratic effects justifiable on methodological grounds alone. However, it is also worth mentioning that such effects are also plausible in the IPTS for substantive reasons. As an example related to perceived burdensomeness, we would not expect (a) that an employee losing 10% of her income is at one-tenth the suicide risk as if she were totally unemployed. Likewise, in the case of thwarted belongingness, a child who loses one parent is not likely at exactly half the suicide risk of a child losing two. In these ways, even straightforward clinical intuition could suggest the effects of perceived burdensomeness and thwarted belongingness on ideation are likely non-linear.
Given a broader call for more tests of fundamental IPTS predictions (Van Orden, 2015) and the unique risk for confounding from quadratic effects, re-examination of the synergy hypothesis is indicated. This investigation tested the interaction between thwarted belongingness and perceived burdensomeness in a sample of university undergraduates and a sample of psychiatric inpatients, incorporating previously neglected quadratic effects of perceived burdensomeness and thwarted belongingness in the model. Prior to analysis, we predicted (a) that a significant and positive interaction between perceived burdensomeness and thwarted belongingness would be detected, both before and after the inclusion of quadratic effects, (b) that there would be significant quadratic effects of both perceived burdensomeness and thwarted belongingness, and (c) that those quadratic effects would manifest curves that are monotonically increasing within the range of the observed data, consistent with the IPTS prediction that increases in perceived burdensomeness and thwarted belongingness should always result in increased risk for suicide ideation (Van Orden et al., 2010).
Section snippets
Sample 1
Sample 1 included 609 undergraduates (79.80% female) from a large university in the Southeastern United States. Following university institutional review board approval, participants who were at least 19 years old were recruited over two semesters using the university׳s online subject pool. After providing informed consent, participants completed a battery of online questionnaires and were given course credit as compensation for participation.
The mean age of Sample 1 was 20.25 (S.D.=1.60;
Results
Descriptive statistics and zero-order correlations are available in Table 1, Table 2. Predictably, the inpatient sample had more severe psychopathology, with 49.11% endorsing some degree of current suicide ideation and 55.37% endorsing at least one prior suicide attempt. However, consistent with previous estimates for college students (Garlow et al., 2008), a notable degree of suicidal behavior was also observed in the undergraduate sample, with 14.57% of participants endorsing some degree of
Discussion
The synergy hypothesis of the IPTS argues two proximal risk factors for suicide ideation (i.e., perceived burdensomeness, thwarted belongingness) are positively interactive, each amplifying the effect of the other. This investigation is part of a broader call for more direct tests of fundamental IPTS hypotheses (Van Orden, 2015) and extends previous work by assessing for quadratic effects, which can distort the size, sign, and significance interaction terms if left out of a model. Taken
References (46)
- et al.
Screening for depression in well older adults: evaluation of a short form of the CES-D
American Journal of Preventive Medicine
(1994) - et al.
Examining the role of emotion in suicidality: negative urgency as an amplifier of the relationship between components of the interpersonal-psychological theory of suicidal behavior and lifetime number of suicide attempts
Journal of Affective Disorders
(2011) - et al.
Psychometric properties of the Beck depression inventory: twenty-five years of evaluation
Clinical Psychology Review
(1988) - et al.
A preliminary test of the interpersonal-psychological theory of suicidal behavior in a military sample
Personality and Individual Differences
(2010) - et al.
Parenting behavior and the interpersonal-psychological theory of suicide: a mediated moderation analysis with adolescents
Journal of Affective Disorders
(2013) - et al.
A test of the interpersonal theory of suicide in a large community-based cohort
Journal of Affective Disorders
(2013) Interaction, nonlinearity, and multicollinearity: implications for multiple-regression
Journal of Management
(1993)Nonlinearity, multicollinearity and the probability of type II error in detecting interaction
Journal of Management
(1998)- et al.
A brief screening tool for suicidal symptoms in adolescents and young adults in general health settings: reliability and validity data from the Australian National General Practice Youth Suicide Prevention Project
Behaviour Research and Therapy
(2002) - et al.
Increase in state suicide rates in the USA during economic recession
Lancet
(2012)