Abstract
Across two countries and two languages, this research examined the multidimensional associations of suicide behaviors (i.e., life-time attempts, life-time communication of intent to others, life-time self-harming, life-time suicide notes, and current suicide ideation) and empirically relevant psychological risk factors (i.e., different facets of mental pain, perceived burdensomeness, thwarted belongingness, and acquired capability), controlling for depressive symptoms. For the Portuguese sample, two underlying dimensions emerged: an ideation dimension and a behavioral dimension, and for the Canadian sample, three dimensions emerged: an ideation dimension and two behavioral dimensions that can be viewed as a splitting of the Portuguese second dimension. Results highlight possible cultural differences between the two countries and that suicide behaviors should be viewed as a multidimensional phenomenon not as a one-dimensional continuum.
Worldwide, death by suicide claims approximately 800,000 lives annually (World Health Organization, 2017). Furthermore, for every suicide, there are many times more occurrences of suicide attempts (World Health Organization, 2014) and suicide ideation. Predicting suicide behaviors remains a very complex and difficult task (Overholser, Braden, & Dieter, 2012), and according to a recent meta-analysis (Franklin et al., 2017), predicting suicidal behaviors has not improved in the last 50 years. In Portugal, the rate of death by suicide was 10.3 per 100,000 in the population in 2017 (National Institute of Statistics, 2019), which represents a similar rate compared with other countries in Europe. This number may, however, reflect an underreporting of suicide cases. In Portugal, there is a national plan of suicide prevention similar to other European countries, but its real implementation in the field has not been fully realized. A few suicide prevention programs for adolescents have been implemented in the last years, such as the program “Mais Contigo” (Santos et al., 2014), but to our knowledge, no specific programs for college students and young adults exist. For Canada, in 2017, the rate for death by suicide was 11.3 per 100,000 (Statistics Canada, 2018). As in other countries, a national plan of action has been formulated but is still in the early stages of being implemented.
The lack of progress in understanding the etiology of suicide, through risk assessment, and suicide outcome, through prediction, could be attributable to a number of assumptions, for example, that suicidal behaviors exist along some sort of continuum (Svetcic & De Leo, 2012) that has at its extreme, death by suicide, with steps along this continuum varying from suicide ideation, communication of intention to others, self-harming, writing a suicide note, to suicide attempts and death by suicide, with similar predictors for each. Klonsky, May, and Safer (2016), however, have emphasized important differences among suicide, suicide attempts, and suicide ideation, indicating that these are distinct phenomena that have different predictors. Several studies (Lewitzka et al., 2017; Mars et al., 2019) have attempted to distinguish between individuals presenting suicidal thoughts and those who have actually attempted suicide, using several types of predictors, in community and clinical populations. For example, Holden and coworkers (Holden & Kroner, 2003; Johns & Holden, 1997) demonstrated the differentiation of suicidal manifestations into distinct dimensions of action orientation and negative cognitions. More recently, different ideation-to-action theories of suicide have been proposed, such as the Interpersonal Psychological Theory of Suicide (Joiner, 2005) or the Integrated Motivational–Volitional Model (Dhingra, Boduszek, & O’Connor, 2015), and these all agree that, first, there is the development of suicidal motivation and, subsequently, there is a progression to attempts, with these being distinct processes having different explanations (see Klonsky, Saffer, & Craig, 2018).
The Interpersonal Theory of Suicide (Joiner, 2005) posits that the unmet interpersonal needs of perceived burdensomeness and thwarted belongingness are necessary for suicide motivation, but that an additional acquired capability for suicide is critical for death by suicide to occur. Other perspectives have also been proposed in order to understand suicidal behaviors. For example, Shneidman (1993) has proposed that suicide is caused by unbearable psychological pain, termed by him “psychache,” with other important risk factors, such as depression and hopelessness only being important through the effect of psychache. Yet another recent theory by Li et al. (2014) proposes a multidimensional psychological pain model of suicide that includes three dimensions: affective, cognitive, and avoidance. Avoidance is regarded as the most important dimension for more serious suicide behaviors and is defined as the tendency to consider suicide as the way to escape from psychological pain. Escaping from mental pain has empirically been demonstrated to be an important motivation for attempting death by suicide (Holden, Kerr, Mendonca, & Velamoor, 1998; Levinger & Holden, 2014). Orbach, Giboa-Schechtman, Johan, and Mikulincer (2004) proposed three dimensions related to tolerance in the face of psychological pain: the ability to tolerate pain surfeit that is the ability to tolerate the excess of mental pain; the ability to cope with mental pain that is a belief that the individual can manage the pain experience; and the ability to contain mental pain, which involves the ability to self-regulate negative emotions. Although these newer theories of suicide show promise, few studies have investigated their contemporaneous contribution to suicidal behaviors, and none, to our knowledge, has investigated their simultaneous contribution to different forms of suicidal behaviors in samples from different countries and cultures.
The aim of this study was to evaluate in two samples from two different countries, Portugal and Canada, the multidimensional relationships between empirically relevant psychological risk factors, including different facets of psychological pain, unmet interpersonal needs, and the acquired capability for suicide; and five different types of suicidal behaviors, not only suicidal ideation and suicide attempts as in previous studies. As such, we assessed current suicidal ideation, life-time communication of suicide intention to others, life-time self-harming behaviors, life-time suicide notes, and life-time suicide attempts. In doing this, we controlled for depressive symptoms, an empirically important risk factor for suicidal behaviors, even in nonclinical populations (Lamis, Malone, Langhinrichsen-Rohling, & Ellis, 2010). We were interested in studying the contribution of other risk factors beyond depressive symptoms. In considering those risk factors, we simultaneously assessed a more internal dimension of risk, that is, psychological pain and its different facets, that has been demonstrated to be a core aspect of risk (see Verrocchio et al., 2016) and also a more interpersonal dimension that included Joiner’s interpersonal theory concepts. This interpersonal theory is one of the most robust and compelling current perspectives of suicidal behaviors (Joiner, 2005). These two risk factors and perspectives may complement each other in accounting for the dimensionality of suicidal behaviors. Within the scope of psychological factors for suicidal behaviors, it is important to ensure that not only intrapsychic but also interpersonal factors are considered.
In keeping with previous research (Holden & Kroner, 2003; Johns & Holden, 1997), it was hypothesized that (a) suicide indicators and risk factors would be associated multidimensionally, that is, more than one dimension comprised of risk factors and suicide indicators would emerge; (b) these associations would be represented by dimensions of ideation and of behaviors; and (c) the nature of the obtained multidimensional structure would replicate across Portuguese and Canadian samples.
Method
Participants and Procedure
Sample 1 consisted of 384 undergraduates (193 women and 191 men) at a midsize Portuguese university (>95% Caucasian) who participated voluntarily without course credit or remuneration. These participants had a mean age of 19.62 years (SD = 2.20, Median = 19.00). Sample 2 consisted of 247 undergraduates (219 women and 28 men) at a midsize Canadian university (>80% Caucasian) who were recruited from an introductory psychology subject pool and who received course credit for their participation. Participants had a mean age of 20.06 years (SD = 5.69; Median = 18.00). The only exclusion criterion was being more than 30 years old. This criterion was set because we were interested in focusing on young adults. Response rate was 96.5% for the Portuguese sample and 99.2% for the Canadian sample.
Individuals in each sample provided informed consent prior to participation. Sample 2 completed the materials online. Sample 1 completed material in classrooms in groups of up to 30. Participants in each sample received information of telephone numbers regarding available counselling resources for participants who wished to talk with a mental health professional. The study was approved by the research ethic review boards of each university.
Materials
Psychological risk factors
Center for Epidemiologic Studies Depression Scale
The Center for Epidemiologic Studies Depression Scale (Radloff, 1977) is a 20-item measure of the frequency of depressive symptoms in the previous 7 days. Items (e.g., “I felt depressed”) are responded on 4-point ratings varying from 0 = never or very rarely–less than 1 day to 3 = very frequently or always–5–7 days. Radloff (1977) has indicated scale score coefficients α of .85 and .90 for community and clinical samples, respectively. For scale score validity, Erford, Johnson, and Bardoshi (2016) have reported a .72 correlation between Beck Depression Inventory–II and Center for Epidemiologic Studies Depression Scale scores across 11 studies involving 3,209 individuals. For the Portuguese sample, we used a previously developed Portuguese version (see Gonçalves & Fagulha, 2004) of the scale. In this study, coefficient α was .91 for the Portuguese sample and .93 for the Canadian sample. The mean score was 14.35 (SD = 10.02) for the Portuguese sample and 21.29 (SD = 12.53) for the Canadian sample.
Interpersonal Needs Questionnaire
The Interpersonal Needs Questionnaire (Van Orden, Cukrowicz, Witte, & Joiner, 2012) assesses current beliefs regarding the degree to which persons feel that their basic need to belong is unmet (i.e., thwarted belongingness) and the degree to which they feel like a burden to others in their lives (i.e., perceived burdensomeness). It includes 15 items, each answered on a 7-point rating varying from 1 (not at all true for me) to 7 (very true for me). Six items assess perceived burdensomeness (e.g., “These days I think I make things worse for the people in my life”) and nine items assess thwarted belongingness (e.g., “These days, I often feel like an outsider in social gatherings”). Hill et al. (2015) have reported strong internal consistency for scores on the Thwarted Belongingness (coefficients α from .81 to .87) and Perceived Burdensomeness (coefficients α from .85 to .90) scales across various samples and, for validity, have reported that scale scores predicted concurrent suicidal ideation across the majority of samples analyzed. For the Portuguese sample, we used a previously developed Portuguese version of the scale (see Costa, Campos, Simões, & Pio, in press). In this study, coefficient α reliabilities were .92 for Perceived Burdensomeness and .82 for Thwarted Belongingness scale scores in the Portuguese sample. Corresponding values were .95 and .92, respectively, in the Canadian sample. In the Portuguese sample, the mean score was 1.39 (SD = 0.72) for the Perceived Burdensomeness scale and 2.37 (SD = 0.99) for the Thwarted Belongingness scale. In the Canadian sample, respective means were 1.90 (SD = 1.24) and 2.91 (SD = 1.33).
Acquired Capability for Suicide Scale
The Acquired Capability for Suicide Scale (Ribeiro et al., 2014) is a scale that specifically measures fearlessness about death (e.g., “I am not at all afraid to die”). Items are answered on 5-point ratings ranging from not at all like me (0) to very much like me (4). Ribeiro et al. have reported scale score coefficient α reliabilities over .80 and, as evidence for validity, a correlation of .37 with intent for the most lethal suicide attempt for psychiatric inpatients. Here, for the Portuguese sample, we used a Portuguese translation of the scale. In addition to the first author, three other clinical psychologists and a bilingual translator, who performed a back-translation, participated in the translation process. In this study, scale score coefficient α was .71 for the Portuguese sample and .83 for the Canadian sample. The mean score was 14.18 (SD = 6.10) for the Portuguese sample and 12.30 (SD = 6.92) for the Canadian sample.
Three-Dimensional Psychological Pain Scale
The Three-Dimensional Psychological Pain Scale (Li et al., 2014) is a 17-item inventory that measures the extent of psychological pain within three aspects: cognitive (e.g., “Whenever I think of my serious shortcomings, I feel a great deal of pain”), affective (e.g., “My pain is emotional rather than physical”), and avoidance (e.g., “I have almost killed myself to make the pain go away”). Items are answered on 5-point Likert ratings that vary from not at all to extremely so. In research with college students, Li, Fu, Zou, and Cui (2017) have reported scale score coefficient α reliabilities over .76 and indicated that the Three-Dimensional Psychological Pain Scale could distinguish among suicide ideators, major depressive disorder patients, and healthy controls. For the Portuguese sample, we used the previously developed Portuguese version of the scale (see Campos, Simões, Costa, Pio, & Holden, 2019). In this study, scale score coefficient α reliabilities in the Portuguese sample were .90, .87, and .84 for the cognitive, affective, and avoidance scales, respectively. Corresponding values were .90, .89, and .88, respectively, in the Canadian sample. In the Portuguese sample, the mean score was 15.99 (SD = 6.55) for the cognitive scale, 12.24 (SD = 6.02) for the affective scale, and 3.70 (SD = 1.86) for the avoidance scale. In the Canadian sample, corresponding values were 18.05 (SD = 7.57), 14.12 (SD = 6.39), and 4.31 (SD = 2.61), respectively.
Psychache Scale
The Psychache Scale (Holden, Mehta, Cunningham, & McLeod, 2001) is a 13-item scale measuring psychological pain (e.g., “I seem to ache inside”). Items are responded on 5-point ratings from 1 (never or strongly disagree) to 5 (always or strongly agree). Psychache Scale scores have shown strong reliability in university (α = .94; Troister & Holden, 2010) and offender (α = .95; Mills, Green, & Reddon, 2005) samples. Scale score validity has been demonstrated by correlations with measures of suicide ideation (r = .65), suicide attempts (r = .45), likelihood of future suicide commission (r = .33), and self-injury (r = .50) (Holden et al., 2001). For the Portuguese sample, we used the previously developed Portuguese version of the scale (Campos, Holden, & Gomes, 2018). In this study, scale score coefficient α values of .93 and .96 were found for the Portuguese and Canadian samples, respectively. The mean score was 23.37 (SD = 9.70) for the Portuguese sample and 25.78 (SD = 11.19) for the Canadian sample.
Tolerance for Mental Pain Scale
The Tolerance for Mental Pain Scale (Orbach et al., 2004) is a self-report measure that evaluates the degree of tolerance to mental pain along three dimensions associated with the model of Orbach et al. (2004): Ability to tolerate pain surfeit, Ability to cope with mental pain, and the Ability to contain mental pain. This scale is composed of 20 items answered on 5-point Likert ratings varying from “1 = not true” to “5 = very true”. Example items are “I cannot concentrate because of my pain.” (reverse scored) for Ability to tolerate scale, “I believe that if I do the right thing, the pain will disappear” for the Ability to cope scale, and “I cannot contain the pain inside me” (reverse scored) for the Ability to contain scale. Higher scores indicate greater ability. For the Portuguese version, we used a Portuguese translation of the scale. In addition to the first author, three other clinical psychologists and a bilingual translator, who performed a back-translation, participated in the translation process. In this study, Cronbach’s α values for tolerate, cope, and contain scales were .92, .75, and .54, respectively, in the Portuguese sample. In the Canadian sample, corresponding values were .93, .69, and .69, respectively. In the Portuguese sample, the mean score for the tolerate scale was 34.86 (SD = 8.23) and 26.38 (SD = 4.30) for the cope scale and 11.37 (SD = 2.43) for the contain scale. In the Canadian sample, corresponding values were 37.25 (SD = 9.49), 24.49 (SD = 5.11), and 12.02 (SD = 2.54), respectively.
Suicide criteria
Life-time suicide attempts were assessed using Item 1 of the Suicidal Behaviors Questionnaire-Revised (SBQ-R; Osman et al., 2001): “Have you ever thought about or attempted to kill yourself.” The SBQ-R consists of four multiple-choice items assessing a history of suicide ideation and/or attempts; recent suicide ideation (i.e., during the past year); the communication of suicidal intentions to others; and the likelihood of a future suicide attempt. In several studies, the four items have been used separately as independent indicators of suicide risk (e.g., Campos & Holden, 2015) Total SBQ-R scores have shown acceptable internal consistency reliability across multiple samples (α coefficients from .76 to .87) and have been found to be effective in distinguishing between suicidal and nonsuicidal individuals (Osman et al., 2001). In this study, we used the Portuguese version of the SBQ-R (Campos & Holden, 2019). Item 1 was scored as: 0 = no suicide attempt (originally answered 1 to 3); 1 = suicide attempt with no intention to die (originally answered 4a); and 2 = suicide attempt with intention to die (original answered 4b). In the Portuguese sample, 12 (3.1%) had a previous suicide attempt and, in the Canadian sample, 17 (6.9%) had previously attempted to die by suicide.
Life-time communication to others was assessed using Item 3 of the SBQ-R: “Have you ever told someone that you were going to commit suicide, or that you might do it,” scored 1 to 3, with 1 corresponding to no, 2 as yes, at one time, and 3 yes, more than once. In the Portuguese sample, 52 (13.5%) had a life-time communication to others. In the Canadian sample, the corresponding number was 55 (22.4%).
Life-time self-harming behaviors were assessed with an item requiring multiple responses: At any time during your life, have you deliberately (in other words, of your own will) taken an overdose of medication (in pills or other form) or hurt yourself in any other way (such as cutting yourself, burning yourself or intoxicating yourself with drugs and/or alcohol) in order to deliberately harm yourself?
Life–time suicide notes were assessed with a single item: “Have you ever written a suicide note”. Possible responses were as follows: 0 = no; 1 = yes, once; 2 = yes, more than once. In the Portuguese sample, 22 (5.5%) had written a suicide note. In the Canadian sample, 30 (12.2%) had written a suicide note.
Current suicide ideation was assessed using the Suicide Ideation Scale. The Suicide Ideation Scale consists of 10 items measuring suicidal ideation in the previous 7 days. Items (e.g., “I have been thinking of ways to kill myself”) are scored on 5-point Likert scales from 1 = never or none of the time to 5 = always or a great many times. Luxton, Rudd, Reger, and Gahm (2011) have reported a coefficient α value of .91 for scale scores associated with a clinical sample. For the Portuguese sample, we used a Portuguese version of the scale. In addition to the first author, three other clinical psychologists and a bilingual translator, who performed a back-translation, participated in the translation process (see Campos, Holden, & Lambert, 2019). In this study, the coefficient α was .90 and .94 in the Portuguese and Canadian samples, respectively. The mean score for the Portuguese sample was 11.57 (SD = 4.24) and for the Canadian sample was 12.67 (SD = 5.39).
Data Analytic Strategy
Results were obtained using SPSS, version 21. Pearson product–moment correlations between scales were used to test relationships among individual risk factors and suicide indicators. Subsequently, our choice of canonical correlation as the statistical analytic method was driven by three considerations. First, canonical correlation can establish the dimensionality of relationships in that it solves for mathematically orthogonal linear combinations of predictors and criteria that relate to one another. Second, in uncovering linear combinations of psychological risk factors and suicide criteria that associate, canonical correlation can potentially identify meaningful latent constructs. Third, canonical correlation better allows for the control of Type I error associated with undertaking several multiple regressions that are not mathematically independent of each other. Because assumptions of data distributional normality and homoscedasticity were not viable, bootstrapping (1,000 resamples) of the correlations between canonical variates was used to confirm the statistical significance of the canonical correlations. For interpretation, the canonical loadings were rotated to a varimax criterion. Canonical solutions were compared between the Portuguese and Canadian samples using coefficients of congruence (MacCallum, Widaman, Zhang, & Hong, 1999) where .98 to 1.00 is an excellent match, .92 to .98 is good, .82 to .92 is borderline, .68 to .82 is poor, and below .68 is terrible.
Results
Sex differences existed between the two samples with proportionately more men in the Portuguese sample (50%) than in the Canadian sample (11%), χ2(1, N = 631) = 97.83, p < .001. The two samples did not differ significantly in terms of age, t(629) = 1.36, p = .17.
Scale correlations are presented in Table 1 for both the Portuguese and Canadian samples. Then, we proceeded with the canonical analysis. With 11 psychological risk factors and five suicide criteria, five possible canonical correlations exist. For the Portuguese sample, canonical correlation pools of 5 (Rc = .85, Wilk’s λ = .22, χ2(55, N = 384) = 561.94, p < .001) and 4 (Rc = .36, Wilk’s λ = .81, χ2(40, N = 384) = 77.22, p < .001) were each statistically significant, indicating that there were two significant independent linear combinations of psychological risk factor and suicide criterion associations for the Portuguese sample. The third canonical correlation was Rc = .21, Wilk’s λ = .93, χ2(27, N = 384) = 25.94, p = .522. The fourth canonical correlation was Rc = .13, Wilk’s λ = .98, χ2(16, N = 384) = 8.68, p = .926. The fifth canonical correlation was Rc = .08, Wilk’s λ = .99, χ2(7, N = 384) = 2.63, p = .917. For the Canadian sample, canonical correlation pools of 5, Rc = .86, Wilk’s λ = .19, χ2(55, N = 245) = 469.59, p < .001; 4, Rc = .36., Wilk’s λ = .73., χ2(40, N = 245) = 77.28, p = .001; and 3, Rc = .29, Wilk’s λ = .83, χ2(27, N = 245) = 43.74, p = .025 were each statistically significant, indicating that there were three significant independent linear combinations of psychological risk factors and suicide criterion associations. The fourth canonical correlation was Rc = .26, Wilk’s λ = .91, χ2(16, N = 245) = 22.90, p = .122. The fifth canonical correlation was Rc = .17, Wilk’s λ = .97, χ2(7, N = 245) = 6.87, p = .447.
Correlations Between Study Variables for Both Samples.
Note. Correlations below the diagonal are for the Portuguese sample (N = 384) and above the diagonal are for the Canadian sample (N = 247).
*p < .05. **p < .01. ***p < . 001.
The rotated canonical loadings are reported in Table 2. For the Portuguese sample, suicidal ideation, assessed with the Suicide Ideation Scale, loaded on the first dimension, an ideation dimension, and the other suicidal behaviors loaded on the second dimension, a behavioral dimension, notably including suicide attempts. Regarding psychological risk factors, perceived burdensomeness, thwarted belongingness, psychache, low ability to cope, and low ability to contain loaded on the ideation dimension. The avoidance and cognitive facets of psychological pain loaded on the behavior dimension. As such, the cognitive facet of pain seems to have importance for both latent dimensions. For the Canadian sample, similar to the Portuguese sample, an ideation dimension was obtained, but in contrast, two different behavioral dimensions emerged. The Portuguese behavior dimension appears to bifurcate with suicide attempts and suicide notes loading on one dimension and self-harming behaviors on the other. Communicating of intent to others seems to have little importance in the three-dimensional Canadian solution. However, if a lower threshold than a .50 loading is considered, it can be said that it cross-loads also on both an ideation and one of the behavior dimensions. As in the Portuguese sample, for the Canadian sample, avoidance of pain associates importantly with suicide attempts as did, but differently than for the Portuguese sample, acquired capability. Low tolerance and the affective and cognitive facets of pain related substantially with self-harming behaviors in the Canadian sample.
Varimax-Rotated Canonical Loadings for the Portuguese and Canadian samples.
Note. Loadings with an absolute value over .50 are in bold.
Canonical solutions were then compared between the Canadian and Portuguese samples using coefficients of congruence. One coefficient exceeded the .92 threshold of good, the first Portuguese and the first Canadian linear combinations (congruence coefficient of .96); one coefficient exceeded the .82 threshold of borderline, the second Portuguese and the third Canadian linear combinations (congruence coefficient of .83); and one coefficient was poor (.73), the second Portuguese and the second Canadian linear combinations.
Discussion
According to our results, associations between the different types of suicidal behaviors and psychological risk factors were multidimensional in nature in each of the Canadian and Portuguese samples, supporting out first hypothesis. As such, based on the present findings, the view that there is a single continuum of suicidal behaviors is not tenable. The present results are consistent with earlier findings of a multidimensionality of suicidal behaviors (e.g., Holden & Kroner, 2003; Johns & Holden, 1997), and with more recent reports of differing predictors for death by suicide, suicide attempts, and suicide ideation (Klonsky et al., 2016).
Results only partially support our second and third hypotheses. In the Portuguese sample, two dimensions emerged, an ideation dimension that included just the suicidal ideation criterion, and a behavior dimension that included all the other suicide criteria, communication of intent, suicide notes, self-harming, and suicide attempts. In the Canadian sample, two behavior dimensions emerged, one containing self-harming behaviors and the other containing suicide attempts and suicide notes. Congruence of the first dimension between samples was good and the congruence of the second Portuguese and the third Canadian linear combinations was acceptable, and between the second Portuguese and the second Canadian linear combination was poor.
Unlike the Portuguese sample, for the Canadian sample, self-harming behaviors had different predictors relative to suicide attempts: the experience of psychological pain and an inability to tolerate the pain versus acquired capability and the tendency to avoid pain. Somewhat unexpectedly in the Portuguese sample, acquired capability had no association with an underlying dimension, neither to the ideation dimension nor with the behavioral dimension. It seems that, for the current Portuguese sample, this acquired capability for suicide is not needed. This result may have at least four reasons. First, it should be noted that Joiner’s interpersonal model of suicide that emphasizes the role of acquired capability was developed in a different cultural context and may not be applicable to the Portuguese culture. Second, it may just be a statistical artifact in that in the Portuguese solution, each of self-harming behaviors and suicide attempts appears in the same dimension, in contrast to the Canadian solution. Third, different percentages of men in each sample may have also influenced the obtained canonical loadings. Finally, this may indeed be a cultural-based result, depending for example, on religious factors. In Portugal, Catholicism is very prominent and perceptions about death may be influenced by this religious prominence. Fearlessness about death may not be related with suicidal behaviors but not being afraid of physical pain, another facet of the acquired capability for suicide that was not assessed may be related to suicidal behaviors. The cognitive dimension of pain was also important for behavior, but the affective dimension of psychological pain was not. With regard to communication of intent, results differ between samples, which may reflect important cultural differences. As in previous, independent research (Holden, Campos, Simões, Costa, Pio, & Lambert, 2019) that compared Portuguese and Canadian samples for several psychological risk factors of suicidal ideation, communication of intent to others and suicide attempts exhibited different results between samples. For the Portuguese sample, according to previous regression analysis results, pain avoidance offered an important contribution to communication in addition to internal perturbation, indicating psychological suffering that cannot be contained and is shared with others. However, for the Canadian sample, it was an extra-punitive motivation that was related to communication, suggesting that, in that sample, communication of intent may have a manipulative nature that has the purpose of controlling others.
Regarding the ideation dimension, psychache and perceived burdensomeness demonstrated substantial loadings for each sample as did low ability to contain pain, but avoidance of pain was important only for the Canadian sample, with thwarted belongingness and low ability to cope with pain being important only the Portuguese sample. Interestingly, whereas some previous research has suggested that thwarted belongingness may be more important than perceived burdensomeness for suicidal motivation in the Portuguese culture (Campos, Santos, Piteira, Abreu, & Tavares, 2018), other non-Portuguese studies have demonstrated the greater influence of perceived burdensomeness for suicidal motivation (Bryan, Clemans, & Hernandez, 2012).
Limitations
First, the dimensionality of findings depends on and is limited by the indicator variables that are assessed in the present research. Thus, the obtained multidimensionality may be a lower estimate of the true structure of suicidal behaviors. Replication with other suicide risk factors and criteria will establish the generalizability of present results. Second, participants were nonclinical samples. This study only recruited college students, and thus the sample representativeness is questionable. Furthermore, individuals of other age and other social class may show a different dimensionality. Identifying the structure of suicidal behaviors in samples of psychiatric populations is also a direction for future investigation. Third, only two countries and two languages were considered in this study. Investigation of additional countries and languages in future research will determine the applicability of current results. Fourth, data were collected using self-report. Although this method has merit, it is dependent on participants accurately recalling events and being candid in answering. Replication using other methods (e.g., report of significant others, interviews, etc.) and other scales assessing corresponding constructs are avenues for additional investigation.
Conclusion
In conclusion, it is important to note that, to our knowledge, no previous research has evaluated the contribution of different psychological risk factors simultaneously with the five types of suicidal criteria that were considered in this study. In partially supporting both previous research and theories across two countries and languages, suicidal risk factors and criteria emerged as an at least two-dimensional phenomenon. As such, findings indicate that a view of suicidal risk factors and criteria as a one-dimensional continuum with completed suicide anchoring one extreme is not a tenable perspective. Although two common latent dimensions of suicidal behaviors, Ideation and Behavior, were derived across Canadian and Portuguese samples, in the Canadian sample, the behavior dimension bifurcated with suicide notes and, particularly, suicide attempts, loading on one dimension and self-harming with communications with others on the other. Communication to others seemed to and acquired capability for suicide had differential importance between the two samples. Such differences may reflect sample-specific variations in age, sex, and linguistic or cultural variations. Continued multivariate investigation of risk factors and suicide indices is encouraged for fully articulating the multidimensionality of the suicide phenomenon across different cultures.
Clinical Implications
Despite this investigation focusing on basic research, results can have some important clinical implications because findings suggest that there are independent, distinct pathways to suicide risk or varying levels of risk, and that different constellations of variables may signal these different levels of risk. Clinicians should be aware of this articulation and understand that different patients may be at different levels of risk and that the pathways to such different status may relate to distinct psychological experiences. Clinical assessment of at-risk individuals should contain extensive clinical interviews and also use objective measures that complement information regarding life history and current problems obtained in the interview. Because asking directly about suicidal behaviors can activate defenses in some resistant patients, inquiring about psychological pain and interpersonal needs may be an alternative avenue for obtaining clinically relevant information regarding different levels of suicide risk.
Footnotes
Acknowledgments
The authors thank all research participants in this study and all the research assistants who helped in the collection and organization of data.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was partially supported by Fundação para a Ciência e a Tecnologia (project reference: UID/CED/04312/2019).