Self-Injurious Thoughts and Behaviors Interview–Revised (SITBI-R): Reliability, Validity, and Inter-Informant Agreement in an Adolescent Sample

Abstract

Objective:

The present study sought to evaluate the psychometric properties and inter-informant agreement of the Self-Injurious Thoughts and Behaviors Interview-Revised (SITBI-R) in adolescents and their parents.

Method:

Suicidal and nonsuicidal adolescents from the community (N =206), ages 12–19 years, were administered the SITBI-R during a lab visit. Approximately half of the adolescents’ parents opted to complete a parent assessment, including the SITBI-R, on behalf of their child. Inter-rater reliability, convergent validity, and inter-informant agreement were assessed.

Results:

The SITBI-R exhibited overall excellent inter-rater reliability and good convergent validity in adolescents. Parent-adolescent agreement ranged from fair to poor across most outcomes, with parents tending not to report past self-injurious thoughts and behaviors endorsed by their children.

Conclusions:

The present study suggests that the SITBI-R can be extended for use with adolescents. Future research should evaluate moderators and implications of parent-adolescent disagreement, particularly with regard to suicide risk.

Suicide represents a major public health concern across the lifespan (World Health Organization, 2019). Suicide attempts are typically preceded by a range of other self-injurious thoughts and behaviors (SITBs), including suicidal ideation, suicide planning, and nonsuicidal self-injury (NSSI; Glenn et al., 2017; Kessler et al., 1999). Refining our ability to detect a range of SITBs allows us to identify higher-risk individuals and is therefore an important component of suicide prevention. Given the increase in incidence of these SITBs in adolescence, it is critical to begin monitoring SITBs during this developmental period (Dervic, Brent, & Oquendo, 2008; Nock et al., 2008; Nock et al., 2013). Moreover, adolescents transition quickly from engaging in one SITB to another: suicide planning typically begins 3–6 months after the onset of NSSI (Glenn et al., 2017), and precedes suicide attempts by just one year (Nock et al., 2013). Adolescence thus marks an especially critical window for comprehensive assessment of a range of SITBs, ideally enabling intervention prior to the escalation of suicidal behavior.

Over the last several decades, a number of SITB measures have emerged for use in both clinical and research settings. These measures are comprised largely of retrospective self-report questionnaires (e.g., the Suicidal Ideation Questionnaire; SIQ; Reynolds, 1987; Suicidal Ideation Scale; SIS; Rudd et al., 1999), and interviewer-administered assessments (e.g., the Columbia Suicide Severity Rating Scale; C-SSRS; Posner et al., 2011; the Scale for Suicide Ideation; SSI; Beck et al., 1974). Assessments can either be singular in focus, for instance measuring suicidal ideation severity exclusively (e.g., SIQ), or more comprehensive, encompassing a range of SITBs and their characteristics, as in the Self-Injurious Thoughts and Behaviors Interview (SITBI; Nock et al., 2007).

The SITBI assesses history and characteristics of both suicidal and nonsuicidal thoughts and behaviors. The SITBI has been widely used in research settings to assess both inpatient (Auerbach et al., 2015; Cha et al., 2016; Glenn et al., 2017) and community and/or outpatient adolescent samples (Barrocas et al., 2012; Nock et al., 2009; Glenn et al., 2017). Recently, the SITBI was updated to address some of its limitations; despite its strong psychometric properties, the original SITBI assessed the presence and characteristics of a more limited range of SITB outcomes, and relied on single-item assessment of several SITBs. For example, the original SITBI assesses suicidal ideation, plan, gesture, attempt, and NSSI, but did not include assessments of aborted and interrupted suicide attempts, or dangerous behaviors with ambivalent intent. It also did not include multi-item assessment of suicide ideation, plan, and attempt. Interviews that include multiple items assessing a broader range of self-injurious outcomes of varying intentionality and lethality have the potential to yield richer, more clinically descriptive information. The Self-Injurious Thoughts and Behaviors Interview-Revised (SITBI-R; Fox et al., 2020) was designed to address these constraints.

The SITBI-R introduced six areas for improvement that both enhanced the precision of the language used and added more behaviors. The areas include (1) assessing dangerous SITB-related areas (i.e., “tempting fate”); (2) increasing specificity around NSSI (i.e., clarifying which types of behaviors are considered NSSI and adding a measure of NSSI severity); (3) including a broader range of suicidal ideation items (i.e., avoid relying solely on a single item to assess ideation); (4) increasing specificity around suicide planning (i.e., adding questions that parse out method, place, and/or time); (5) differentiating suicide threats (i.e., communications) and gestures (i.e., behaviors); and (6) adding questions that assess suicidal intent of self-injurious behaviors and the urge to act on suicide thoughts and plans. For a more comprehensive overview of the rationale behind each of these additions, see Fox et al. (2020).

Several existing suicide measures, including the SITBI-R, have not been validated for use with adolescents. More specifically, several interviewer-administered assessments and self-report questionnaires were either designed for adult samples (e.g., SSI), or have not yet been assessed for use in adolescent samples (e.g., SITBI-R, SIS). Further, a significant proportion of the measures that are validated for use in adolescents rely on a single module or item (e.g., Kiddie Schedule for Affective Disorders and Schizophrenia; K-SADS; Kaufman et al., 1997; Patient Health Questionnaire-9; PHQ-9; Richardson et al., 2010; Moods and Feelings Questionnaire; MFQ; Wood et al., 1995). Although this is better than not assessing SITBs at all, research suggests that single-item measures of SITBs often fail to fully capture the nature of these phenomena and may lead to misclassification (Millner, Lee, & Nock, 2015). Moreover, these approaches frequently assess just one SITB, such as thoughts of suicide or a wish to be dead, while prior research suggests that suicidal ideation often co-occurs with other, more severe SITBs in adolescence (e.g., suicide planning, suicide attempts, aborted attempts; Glenn et al., 2017). Previous research also suggests that NSSI, a robust predictor of future suicidal behavior, is more likely to emerge in early adolescence, while suicidal thoughts and behaviors are more likely to emerge in later adolescence and young adulthood (Jacobson & Gould, 2007; Kessler et al., 1999; Nock et al., 2012; Whitlock et al., 2011). It is therefore important to assess a wide range of SITBs in adolescents and ensure that comprehensive assessment tools used to assess these SITBs extend reliably from adults to adolescents. Research is thus needed to determine the psychometric properties of more comprehensive SITB instruments, often used in adult samples, for use with adolescents.

Finally, a critical consideration of the assessment of psychopathology in youth is the inclusion of multiple informants, such as parents or teachers, whose reports are often incorporated into clinical formulations to maximize successful evaluation and treatment (Kraemer et al., 2003; Hunsley & Mash, 2007). Multi-informant assessment of adolescents’ SITBs is likely to be an important part of assessment, but is complicated by the fact that adolescents and their parents often share discrepant reports. Recent research suggests that children and adolescents tend to report higher rates of suicidal thoughts and behaviors than do their parents (Breton et al., 2002; Deville et al., 2020; Kashani et al., 1989; Klaus et al., 2009; Sourander et al., 1999). For instance, a recent study found that among adolescent suicide ideators, approximately half of their parents were unaware of their child’s suicidal thinking (Jones et al., 2019). Such discrepancies are observed across a range of youth externalizing and internalizing disorders (Achenbach, McConaughy, & Howell, 1987; De Los Reyes et al., 2015).

While informant discrepancies can introduce challenges, they can also be leveraged to improve assessment validity (De Los Reyes & Kazdin, 2005; Kraemer et al., 2003). For example, previous research suggests that implementing a systematic approach to the integration of reports across multiple informants (e.g., trait-score approach) is associated with improved prediction of clinical outcomes compared to single-informant reports (Makol et al., 2020). Given the challenge clinicians and researchers have historically faced in predicting suicide-related outcomes (Franklin et al., 2017; Wang et al., 2016), multi-informant approaches may be especially promising with regard to SITB assessment. In assessing psychometric properties of the original SITBI among adolescents, Nock and colleagues (2007) found relatively strong parent-child agreement on the presence of suicidal ideation, attempt, and NSSI in a community sample of adolescents. However, the extent of parent-child agreement across a much wider range of SITBs assessed in the SITBI-R has yet to be examined, and is an important component of understanding basic properties of the SITBI-R for use among adolescents.

The present study addresses these knowledge gaps by examining, for the first time among youth, the psychometric properties and inter-informant agreement of the SITBI-R. The SITBI-R is a semi-structured, comprehensive interview that assesses characteristics of SITBs (e.g., presence, frequency, severity) across multiple timeframes (e.g., past week, past year, lifetime) (Fox et al., 2020). We pursued two aims. First, we sought to examine the reliability and validity of the SITBI-R in an adolescent sample. In doing so, we build on recent work by Fox and colleagues (2020), who tested the SITBI-R in adults, to adolescents. Given that the SITBI-R is one of the few instruments to collect a comprehensive history of a range of SITB outcomes, many of which emerge or peak in adolescence, demonstrating its reliability and validity in an adolescent sample is important to establish its research and clinical value among this high-risk population. Second, we sought to explore inter-informant agreement for parallel SITBI-R items completed by adolescents and, separately, by their parents, to understand the extent to which parents might serve as a reliable, supplementary source of information about adolescents’ SITBs.

In line with prior research (e.g., Fox et al., 2020; Jones et al., 2019; Nock et al., 2007), we expected our investigation to yield: (1) excellent inter-rater reliability for presence, frequency, and age of onset for a range of SITB outcomes; (2) good to excellent convergent validity for a range of SITB outcomes; and (3) poor to moderate inter-informant agreement across a range of SITB outcomes between adolescents and their parents, due specifically to parents reporting fewer past SITBs compared to adolescents’ self-reported SITBs.

Methods

Participants

Participants were recruited from the community via street fairs, flyers, and online advertisements (e.g., social media, ResearchMatch) for two lab-based studies examining cognitive mechanisms in suicidal and nonsuicidal adolescents.¹ Potential participants were contacted by post-baccalaureate- and masters-level research staff who conducted phone screens to determine eligibility. For the present investigation, adolescents must have either endorsed suicidal ideation in the past year, or indicated no lifetime history of suicidal ideation or attempt. Additional inclusion criteria were: 12–19 years, English fluency, and ability to complete an in-person interview. Study exclusion criteria were: the presence of any factor impairing the adolescent’s ability to effectively participate in the study, including the presence of gross cognitive impairment, agitated/violent behavior, or high/imminent risk of suicide.

Participants included 210 adolescents, four of whom did not complete the SITBI-R and were therefore removed from analyses. The analytic sample was thus comprised of 206 participants between 12 and 19 years (M =16.86, SD = 1.97; Table 1), the majority of whom identified as female (70%). Participants were recruited from an urban area and were racially diverse, with 33% identifying as White, 22% Black/African American, 25% Asian, and 18% multiracial/other. Nearly one-quarter (24%) of participants reported Hispanic ethnicity. Almost one-third of the sample (32%) identified as a sexual minority (i.e., not heterosexual). Half (n=105) of participants were accompanied by a parent who opted to complete a parent assessment.

Table 1.

Sample Demographics (N=206)

	M(SD)	Min	Max	n	%
Age (years)	16.86 (1.97)	12	19
Gender
Female				113	70.2
Male				42	26.1
Transgender Male				2	1.2
Gender Variant/Non-Conforming				1	0.6
Other				3	1.9
Race
White				66	33.3
Black/African-American				43	21.7
Asian				50	25.3
Other				36	18.2
Unknown				3	1.5
Ethnicity (% Hispanic)				48	24.4
Sexual Orientation
Heterosexual/Straight				129	65.5
Homosexual				13	6.6
Bisexual				33	16.8
Questioning				8	4.1
Other				9	4.6
Unknown				5	2.5

Measures

Self-Injurious Thoughts and Behaviors Interview-Revised.

Adolescents were administered the Self-Injurious Thoughts and Behaviors Interview-Revised (SITBI-R; Fox et al., 2020) by post-baccalaureate, masters, and doctoral level research staff who had been trained and supervised by an experienced clinical psychologist. SITBs assessed during the interview included NSSI (and NSSI method), suicidal ideation (and specific suicidal thoughts), thinking of a suicide method, place, and time, suicide gesture, aborted attempt, interrupted attempt, suicide attempt, tempting fate, and tempting fate with possible suicidal intent, across several timeframes. Interviews with adolescents were audio recorded with participant consent and were completed with research staff in a separate room from accompanying parents. Parents who agreed to participate themselves completed items from the SITBI-R about their child’s SITB engagement, presented in a self-report format. Parents had the option to respond to yes/no questions about their child’s past SITBs with “unsure.”

Suicidal Ideation Questionnaire.

Participants completed the Suicidal Ideation Questionnaire (SIQ; Reynolds, 1987), a 30-item self-report measure rated on a 7-point scale. The SIQ assesses presence and frequency of past-month suicidal ideation and was used to establish convergent validity. While the SIQ primarily assesses ideation over the previous month, two of its anchors (i.e., “I have never had this thought” and “I had this thought but not in the past month”) provide information about lifetime presence, and it was thus possible to use to assess convergent validity across timeframes.

Procedure

Participating adolescents completed a three-hour in-person lab visit. Adolescents under 18 years (n = 92) were required to have a parent (or legal guardian) accompany them to the lab visit; those 18 and older (n = 98) were given the choice to come to the lab unaccompanied. Participants completed a battery of self-report measures, computer tasks, and interviews. Youth with a history of suicidal thoughts and behaviors also received risk assessments and safety plans during the in-person lab visit. Adolescents were compensated with an Amazon gift card and received a roundtrip Metro Card to reimburse travel expenses. Any parent who accompanied adolescents to the in-person lab visit, regardless of the adolescents’ age, was encouraged but not required to complete a self-report questionnaire regarding their child’s mental health, including items from the SITBI-R, to assess for their child’s engagement in SITBs. Self-consenting parents who completed this questionnaire were also compensated with an Amazon gift card and a round trip MetroCard. All study procedures were approved by the Teachers College, Columbia University Institutional Review Board.

Statistical Analysis

The present study was pre-registered as secondary analyses of an existing dataset on the Open Science Framework.² Analyses were conducted in SPSS (version 26; IBM SPSS Inc., Chicago, Illinois).

Sample Descriptives.

Frequencies, means, and standard deviations of lifetime, past year, and past week SITBs, as well as age of onset, were conducted, as appropriate. Descriptive statistics are reported according to adolescents’ reports of past SITBs.

Inter-Rater Reliability.

Inter-rater reliability was assessed by comparing SITBI-R interview responses obtained by interviewers and those interpreted by an independent reviewer and licensed clinical psychologist (CC), who listened to 40 (≈20%) of the original SITBI-R interview recordings (i.e., those conducted by trained interviews) and separately filled out (i.e., coded) new SITBI-R packets based on the audio recordings. Fleiss’ Kappas (Fleiss, Levin, & Paik, 2003), two-way random effects absolute IntraClass Correlation Coefficients (ICCs), and Spearman’s Rho correlations (Dancey & Reidy, 2007) were conducted, as appropriate. Inter-rater reliability analyses assessed a combination of lifetime, past year, and past week presence (dichotomous) and frequency (continuous), as well as age of onset (continuous), for a range of SITB outcomes, including NSSI (and each NSSI method), suicidal ideation (and each type of thought), thinking of a suicide method, place, and time, suicide gesture, aborted attempt, interrupted attempt, suicide attempt, tempting fate, and tempting fate with possible suicidal intent.

Convergent Validity.

Convergent validity between the SITBI-R and the SIQ was assessed with Fleiss’ Kappa because all variables assessed were categorical (Fleiss, Levin, & Paik, 2003). Convergence between measures on lifetime presence of suicidal ideation, thoughts of suicide methods, and thoughts of a suicide time were examined. Additionally, past week presence of suicidal ideation and suicide planning (i.e., thinking of a method, place, or time), as measured by the SITBI-R were compared to past month presence of suicidal ideation and suicide planning (i.e., thinking of a method or time) as measured by the SIQ.³ Of note, timeframes measured across instruments were distinct but overlapping (i.e., SITBI-R assessed past week presence of suicidal ideation and planning; SIQ assessed past month presence of suicidal ideation and planning).

Inter-Informant Agreement.

Inter-informant agreement analyses were conducted to examine convergence of responses between adolescents and parents using Fleiss’ Kappa, ICCs, and Spearman’s Rho, as appropriate (Fleiss, Levin, & Paik, 2003; Nock et al., 2007; Dancey & Reidy, 2007). Agreement was assessed for a combination of lifetime presence (dichotomous)/frequency (continuous), past year presence (dichotomous)/frequency (continuous), past week presence (dichotomous), and age of onset (continuous) for the following variables: NSSI, suicidal ideation, thinking of a suicide method, place, and time, suicide gesture, aborted attempt, interrupted attempt, suicide attempt, and tempting fate. For frequency and age of onset analyses, parent-adolescent dyads were only included when both informants endorsed a behavior and were thus both prompted with an assessment of characteristics. Parent-adolescent dyads in which parents responded with “unsure” to a particular SITB were excluded from analyses involving that variable. To explore the directionality of inter-informant discordance, we calculated percentages of parents who endorsed a behavior that their child endorsed, as well as the percentage of adolescents who endorsed a behavior that their parent either endorsed or responded to with “unsure.”

Interpretation.

For Fleiss’ Kappa analyses, Kappa values range from −1 to 1. Following Nock and colleagues (2007) and Fox and colleagues (2020), we considered values below .40 to suggest poor agreement, values from .40 to .75 to suggest fair to good agreement, and values greater than .75 to suggest excellent agreement (Nock et al., 2007; Fleiss, Levin, & Paik, 2003). While there are not universally accepted values regarding acceptable reliability using ICCs, higher numbers indicate stronger reliability/agreement (Koo & Li, 2016). However, when responses are very similar, for example when there is no or little history of a particular behavior, ICCs tend to be low. We therefore also conducted Spearman correlations as an additional measure. Rho values range from −1 to 1. We considered scores of 1 to suggest perfect reliability, .7-.9 to suggest excellent reliability, .4-.6 to suggest moderate reliability, .1-.3 to suggest weak reliability, and less than or equal to 0 to suggest no reliability (Fox et al., 2020).

Results

Sample Descriptives

With regard to suicidal thoughts, 40% of participants (n = 82) reported having thoughts of killing themselves in their lifetime, with 35% (n = 69) endorsing these thoughts within the past year, and 12% (n = 24) within the past week. The average age of onset of suicidal ideation in our sample was 13 years (SD = 3.06). Thirty percent of participants (n = 61) endorsed thoughts about a specific suicide method, with 12% of participants (n = 25) endorsing thoughts of a specific suicide place and 11% (n = 23) a specific time. Twenty-three percent of participants (n=47) thought about at least one aspect of their suicide plan (i.e., method, place, or time) within the past year and 3% (n=6) within the past week. The average age of onset for suicide planning was 13.7 years (SD = 2.49).

With regard to self-injurious behaviors, 28% of adolescents (n = 58) reported engaging in NSSI in their lifetime, with 19% (n =39) and 2% (n = 4) endorsing having done so within the past year and week, respectively. The average age of onset of NSSI was 13.6 years (SD = 2.49). Eleven percent of participants (n = 23) endorsed a lifetime history of suicide attempt, with 2% (n = 4) endorsing past year history, and none endorsing past week history. Fifteen percent of participants (n = 31) endorsed a lifetime history of aborted suicide attempt, with 10% (n = 20) reporting them in the past year, and one participant in the past week. Finally, 7% (n = 15) of the sample endorsed lifetime histories of an interrupted attempt, with 4% (n = 8) endorsing past year interrupted attempts, and no participants endorsing past week interrupted attempts. Average age of onset for suicide attempt was 14.3 years old (SD = 2.13). Further details regarding lifetime, past year, and past week SITB histories are available in Supplementary Table 1. Clinical and demographic characteristics of the subsamples used to assess inter-rater reliability and inter-informant agreement can be found in Supplementary Tables 2 and 3, respectively.

Inter-Rater Reliability

Results of lifetime inter-rater reliability analyses are presented in Table 2. The SITBI-R exhibited excellent inter-rater reliability (κ = 0.93– 0.94) for lifetime and past year presence of NSSI.⁴ All specific types of NSSI demonstrated perfect inter-rater reliability for lifetime presence (κ = 1.00), with the exception of “cut skin”, “insert sharp objects under the skin” and “scratch self” (κ = −0.03 – 0.53). NSSI lifetime frequency and age of onset also exhibited close to perfect (ρ = 0.99, ICC=1.00, p <.001) and perfect inter-rater reliability (ρ = 1.00, ICC=1.00, p <.01), respectively. Inter-rater reliability was also excellent to perfect (κ = 0.83 – 1.00) for lifetime presence of suicidal ideation (including all specific suicidal ideation thoughts) and good to excellent for past week and past year presence of suicidal ideation (κ = 0.75 – 0.82). Reliability for age of onset and lifetime frequency of suicidal ideation was also perfect (ρ= 1.00, ICC=1.00, p <.01) and excellent (ρ= 0.78, ICC=0.99, p <.001), respectively. The SITBI-R also demonstrated perfect (κ = 1.00) and excellent (κ = 0.94) inter-rater reliability for lifetime and past year presence of suicide plan respectively, though past week presence of suicide plan exhibited fair reliability (κ = 0.64). Lifetime frequency and age of onset of thinking of a suicide plan demonstrated excellent (ρ = 0.80, p = .01, ICC = 0.91, p <.001) and perfect reliability (ρ= 1.00, ICC=1.00, p <.01), respectively. There was also perfect reliability (κ = 1.00) for lifetime and good past year presence of suicide gesture (κ = 0.65). There was perfect reliability (κ = 1.00) for lifetime and past year presence of suicide attempt, aborted attempt and interrupted attempt. Reliability for tempting fate was excellent for lifetime (κ = 0.93) and perfect for past year (κ = 1.00), while reliability for tempting fate with possible suicidal intent exhibited good to excellent reliability (κ = 0.44– 0.86) for lifetime and past year presence.

Table 2.

Inter-Rater Reliability

	Lifetime Presence κ (n)
Nonsuicidal Self-Injury	0.94 (38)
Cut skin	0.53 (37)
Hit self	1.00 (38)
Burned skin	1.00 (38)
Insert sharp objects	−0.03 (38)
Scratch skin	0.47 (38)
Suicide Ideation	1.00 (38)
Seriously considered killing self	1.00 (37)
I wish I could disappear or not exist	0.94 (38)
I wish I were never born	0.94 (38)
Life is not worth living	0.83 (38)
I wish I could go to sleep and never wake up	0.94 (38)
I wish I were dead	1.00 (38)
Maybe I should kill myself	0.94 (38)
I should kill myself	1.00 (38)
I am going to kill myself	1.00 (38)
Suicide Plan: Method	1.00 (37)
Suicide Plan: Place	0.91 (37)
Suicide Plan: Time	0.89 (36)
Suicide Gesture	1.00 (38)
Aborted Attempt	1.00 (37)
Interrupted Attempt	1.00 (38)
Suicide Attempt	1.00 (40)
Tempting Fate	0.93 (39)
Tempting Fate with possible suicidal intent	0.86 (24)

Note. All ratings were the same for “bit self.”

Convergent Validity

Convergent validity ranged from good to excellent across all investigated variables (i.e., suicidal ideation, suicide plan). There was excellent agreement across the SITBI-R and SIQ regarding lifetime presence of suicidal ideation (κ = 0.84) and good agreement across measures regarding presence of past week/month suicidal ideation (κ = 0.59). Similarly, agreement was excellent for lifetime thoughts about a specific suicide method (κ = 0.82) and good for lifetime thoughts about a specific suicide time (κ = 0.51). Finally, agreement was good for recent (i.e., past week/past month) thoughts of suicide planning (κ = 0.66).

Inter-Informant Agreement

Clinical and demographic differences between participants included in inter-informant agreement analyses (i.e., those who had a parent complete the parent survey) and those not included are presented in Supplementary Table 4. Participants included in inter-informant agreement analyses were slightly younger in age on average (M = 15.46 years) compared to those not included (M = 18.26 years), p < .001. Participants included in inter-informant agreement analyses were also slightly less likely to have a past suicide attempt (6.8%) compared to those not included (15.8%), χ2 = 4.17, p < .05.

Results of inter-informant agreement analyses are presented in Table 3. Parent-adolescent agreement was fair (κ = 0.49 – 0.63) for lifetime, past year, and past week presence of NSSI. For parent-adolescent dyads in which both parent and child reported lifetime history of NSSI (n =13), agreement regarding past year frequency was fair (ρ= 0.55, ICC=0.17, p > .05), and age of onset was poor (ρ= 0.08, ICC=0.34, p >.05).

Table 3.

Inter-informant Agreement

	Lifetime Presence	Past Year Presence	Past Week Presence
	κ (n)	κ (n)	κ (n)
Nonsuicidal Self-Injury	0.63 (99)	0.62 (99)	0.49 (99)
Suicide Ideation	0.59 (84)	.23 (75)	0.37 (75)
Suicide Plan: Method	0.42 (90)	.	.
Suicide Plan: Place	0.36 (86)	.	.
Suicide Plan: Time	0.31 (84)	.	.
Thinking of a Suicide Plan	.	0.40 (87)	−0.01 (87)
Suicide Gesture	0.01 (98)	.	.
Aborted Attempt	−.04 (94)	−0.03 (94)	.
Interrupted Attempt	−.03 (95)	−0.01 (95)	.
Suicide Attempt	0.37 (100)	.	.
Tempting Fate	0.13 (99)	.	.

Note. Cells with missing data reflect SITBs that were either not assessed or did not have a sufficient number of adolescents endorsing them to calculate agreement.

With regards to suicidal ideation, agreement ranged from fair to poor (κ = 0.23 – 0.59) for lifetime, past year, and past week presence, with past year presence having the lowest agreement. Among parent-adolescent dyads in which both informants endorsed lifetime history of suicidal ideation (n = 12), agreement regarding past year frequency of suicidal ideation was excellent (ρ= 0.91, p <.01, ICC=0.23, p > .05), and age of onset was fair (ρ= 0.42, ICC=0.42, p <.05).

Agreement regarding lifetime history of thinking of a suicide method, place, and time ranged from fair to poor (κ = 0.31 – 0.42), as did past week and past year presence of thinking about a suicide plan (i.e., suicide method, place, or time; κ = −0.01 – 0.40). Lifetime and past year presence of aborted attempts and interrupted attempts both yielded poor agreement (κ = −0.04 - −0.01), as did lifetime presence of tempting fate and suicide gesture (κ = 0.01 – 0.13). Parent-adolescent agreement regarding lifetime history of a suicide attempt was poor (κ = 0.37).⁵

An overview of the directionality of parent-adolescent disagreement on lifetime SITB history is presented in Figure 1. Among adolescents who endorsed a lifetime history of suicidal ideation, thinking of a suicide method, place, time, suicide gesture, aborted attempt, interrupted attempt, suicide attempt, and tempting fate, the majority of parents reported either that they were unsure or that their child had not engaged in the SITB. Parents appeared most aware of NSSI reported by their children, with 55% of those whose children endorsed NSSI reporting the same, followed by suicidal ideation (43%), suicide method (32%), and suicide attempt (29%).

Figure 1.

Parent responses when adolescents endorse self-injurious thoughts and behaviors

Note. Adolescent-parent dyads in which adolescents endorse a given thought/behavior had the following n’s per variable: NSSI=20, suicidal ideation=35, suicide method = 25, suicide place=9, suicide time=8, suicide gesture=9, aborted attempt=13, interrupted attempt=6, suicide attempt=7, and tempting fate= 9.

Among parents who reported lifetime histories of specific SITBs on behalf of their child, most often, adolescents’ ratings coincided. Specifically, among parents reporting NSSI histories in their children, 79% of adolescents agreed, with 88% agreeing for suicidal ideation, 80% for suicide method, 67% for suicide place, 100% for suicide time, and 67% for suicide attempt. Notable exceptions include parent report of suicide gesture, in which only 10% of their children’s ratings agreed, aborted and interrupted attempts, with 0% accuracy, and tempting fate, with 33% accuracy. A breakdown of adolescent responses when parents report uncertainty about a particular SITB is presented in Figure 2. When parents reported that they were unsure whether or not their child engaged in a particular SITB, most of their children did not report engaging in that SITB.

Figure 2.

Adolescent responses when parents report “unsure” to self-injurious thoughts and behaviors

Note. Adolescent-parent dyads in which parents report unsure to a given thought/behavior had the following n’s per variable: NSSI=5, suicidal ideation=20, suicide method = 13, suicide place=17, suicide time=19, suicide gesture=5, aborted attempt=9, interrupted attempt=8, suicide attempt=3, and tempting fate= 4.

Discussion

The present study is the first to assess psychometric properties of the SITBI-R among adolescents. Findings were largely consistent with our hypotheses: the SITBI-R demonstrated mostly excellent inter-rater reliability and strong construct validity overall, with few exceptions. Inter-informant agreement, as expected, was largely fair to poor between adolescents and their parents. Notably, discrepancies revealed that parents tended not to report STIBs endorsed by their children.

Our findings supporting mostly excellent inter-rater reliability of the SITBI-R are in line with prior studies assessing the original SITBI in adolescents (Nock et al., 2007), as well as the SITBI-R and the Spanish version of the SITBI in adults (Fox et al., 2020; García-Nieto, Blasco-Fontecilla, Paz Yepes, & Baca-García, 2013). Still, the present study is the first to offer support for the inter-rater reliability of the newer features for use with adolescents, including aborted and interrupted attempts, tempting fate, specific methods of NSSI, and specific suicidal thoughts. Our results demonstrated excellent to perfect agreement among raters for the vast majority of SITBs assessed with a few exceptions—described below.

Regarding lifetime presence of various SITB outcomes, three specific methods of NSSI were notable exceptions to otherwise excellent inter-rater reliability; reliability was fair for lifetime presence of scratching skin and cutting skin, and poor for lifetime presence of inserting sharp objects into skin or nails. One possible explanation for the inter-rater disagreement is variation in raters’ interpretations. The question assessing NSSI method is open-ended rather than forced choice, leaving room for interviewer interpretation in the classification of the method employed. Indeed, in our study, what one interviewer coded as inserting sharp objects into skin or nails, another coded as scratching skin, resulting in lower inter-rater reliability for both items. Other exceptions were in the case of past week suicide plan, which includes thinking about a method, place, or time for suicide, and past year presence of tempting fate when unclear if one wanted to live or die, where reliability was fair. Fair reliability in these instances might be due to discrepancies in either interviewee or interviewer interpretation, in this case regarding different possible iterations of suicide planning, or of what constitutes possible suicidal intent when tempting fate. Difficulties in interviewer interpretation may parallel participants’ uncertainty about their own suicidal intent, which is not always clear. Nonetheless, the overall excellent inter-rater reliability of the SITBI-R among our sample suggests this interview can be reliably extended for use with adolescents.

Also consistent with our hypotheses, our findings indicate overall strong convergent validity between the SITBI-R and a widely used self-report questionnaire of suicidal ideation in adolescents. These results are in line with extant literature supporting adequate construct validity of the original SITBI and German SITBI in adolescent populations (Fischer et al., 2014; Nock et al., 2007), and the SITBI-R in adult populations (Fox et al., 2020). Specifically, we found excellent agreement between both instruments for lifetime suicidal ideation and lifetime thoughts of a suicide method, and good agreement for lifetime thoughts of a suicide time, past week/past month suicidal ideation, and past week/past month thinking of a suicide plan, despite slight inconsistencies in definitions and timeframes in the measures used. Taken together, the nuanced differences between measures may explain why agreement on past week/month variables were lower than agreement on lifetime presence of suicidal thoughts and thoughts of a suicide method—variables devoid of discrepancies across measures.

Finally, our examination of parents as informants of their adolescents’ SITBs yielded poor to fair agreement for the vast majority of SITBs assessed, as expected. The present study is the first to our knowledge to assess parent-adolescent agreement across a wide range of SITBs beyond suicidal ideation, plan, attempt, and NSSI, and including, for instance, aborted and interrupted attempts, tempting fate, and suicide gestures. Disagreement in lifetime history across SITBs was primarily driven by adolescents reporting SITBs not reported by their parents, with suicide gesture being a notable exception; of the parents who reported suicide gestures in their children, only 10% of adolescents’ reports agreed. Unsurprisingly, some of the SITBs least likely to be recognized by parents included more specific thoughts and behaviors, perhaps less likely to be discussed with parents, such as thinking of a suicide place and time, engaging in an aborted or interrupted attempt, and tempting fate. Among adolescents who endorsed aborted and interrupted attempts, as well as tempting fate, more than half of their parents did not report these behaviors. In contrast, parents appeared to endorse NSSI, suicidal ideation, suicide attempt, and thoughts of a suicide method at higher rates – though rates of agreement for these SITBs were still low.

These findings indicating overall low levels of parent-child agreement on SITBs are consistent with much of the existing literature (Jones et al., 2019; Klaus et al., 2009; Thompson et al., 2006). In the largest study of parent-adolescent agreement of suicidal thinking to date (Jones et al., 2019), 55% of parents did not report that their child experienced thoughts of killing themselves when their children reported they had. This low rate of agreement is comparable to the 57% of parents in our study who either were unsure or believed their child did not have thoughts of suicide, when in fact they reported the opposite. However, in contrast to previous studies, which tend to find stronger agreement for suicidal behaviors than thoughts (Klaus et al., 2009), and more broadly for externalizing rather than internalizing problems (De Los Reyes et al., 2015), parent-child agreement in our sample was weaker for suicide attempt than suicidal ideation. In fact, less than one third of the parents of adolescents who reported a lifetime history of suicide attempt in our study reported their children’s attempt histories. One potential explanation is that suicide attempts experienced within our sample were less lethal and thereby less likely to require third-party response (e.g., parental involvement). This is possible as our sample was community-based, in contrast to Klaus and colleague’s (2009) assessment of psychiatrically hospitalized adolescents.

Interestingly, for lifetime history of the majority of SITBs, more often than not, children of parents who responded “unsure” did not report engaging in that SITB. The ability for parents in our study to respond with “unsure” to each SITB assessed allowed a more nuanced understanding of parents’ levels of awareness. Given that most often parents reported unsure when their children did not in fact endorse that SITB, it seems plausible that parents in our sample who made this choice did so out of an abundance of caution. Additionally, the fact that all parents had the option to respond with uncertainty, and yet many often made a dichotomous yes/no choice, suggests that parents who respond negatively when their child responds affirmatively genuinely believe their child has not engaged in the SITB. The present study builds upon the growing body of literature suggesting that multi-informant assessment of SITBs have the ability to paint a richer clinical picture of an adolescent’s experiences (Jones et al., 2019; Klaus et al., 2009). Ultimately, more work is needed to understand the implications of parent-child discordance, and how discrepant reporting can ultimately be harnessed in service of the safety and well-being of youth with SITBs.

Limitations, Future Directions, and Conclusion

Our findings should be considered in light of several limitations. First, our sample was recruited from the community, and while this enabled recruitment of a diverse sample, it included adolescents both with and without histories of engaging in SITBs. Relatedly, adolescents’ participation in this study may have necessitated parental knowledge in some cases, which may have influenced our findings pertaining to inter-informant agreement. Future research should examine the SITBI-R for use among more clinically severe samples. Such an approach would also likely yield larger numbers of respondents endorsing outcomes, enabling analyses of SITBs we were unable to examine due to an insufficient number of participants endorsing certain outcomes. Second, and relatedly, due to all participants responding the same way to certain items, there were some variables for which we were unable to examine inter-rater reliability, especially those that inquire about the past week. Third, the instrument used to determine convergent validity (i.e., SIQ) did not map on perfectly to timeframes and definitions used in the SITBI-R, which may have influenced results. Future research should seek to expand our understanding of variables we could not assess and of the variables that did not yield excellent reliability or validity, including tempting fate with possible suicidal intent, specific methods of NSSI, and suicide planning, while also assessing additional forms of reliability we were unable to examine, including test retest reliability and internal consistency. Fourth, observed differences in Kappa values may have been influenced by variable base rates of SITB outcomes (Feinstein & Cicchetti, 1990). Finally, more work is needed to understand the factors that influence inter-informant agreement between adolescents and their parents. A deeper understanding of the nature of the discordance – for example, moderators of disagreement – and how discrepant reports of SITBs can best be integrated would have important implications for the assessment and prevention of SITBs in youth.

In sum, the present study provides support for the SITBI-R as a reliable and valid measure of SITBs among a community sample of adolescents, suggesting promise for its continued use across clinical and research settings.