Establishing the psychometric properties and construct validity of the Painful and Provocative Events Scale-Revised

Accepted Manuscript

Establishing the Psychometric Properties and Construct Validity of the Painful and Provocative Events Scale–Revised Lauren N. Forrest , Elizabeth A. Velkoff , Courtney J. Johnson , Aaron Luebbe , April R. Smith PII: DOI: Reference:

S0165-0327(19)30358-1 https://doi.org/10.1016/j.jad.2019.04.074 JAD 10735

To appear in:

Journal of Affective Disorders

Received date: Revised date: Accepted date:

9 February 2019 5 April 2019 17 April 2019

Please cite this article as: Lauren N. Forrest , Elizabeth A. Velkoff , Courtney J. Johnson , Aaron Luebbe , April R. Smith , Establishing the Psychometric Properties and Construct Validity of the Painful and Provocative Events Scale–Revised, Journal of Affective Disorders (2019), doi: https://doi.org/10.1016/j.jad.2019.04.074

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE Highlights The current measure of painful and provocative events is not psychometrically valid Exploratory factor analysis identified a new painful and provocative events scale Confirmatory factor analysis confirmed that the new scale replicates The revised scale has a reliable and replicable factor structure The psychometrically sound scale can improve the study of suicidal behavior

AC

CE

PT

ED

M

AN US

CR IP T

    

1

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

2

Establishing the Psychometric Properties and Construct Validity of the Painful and Provocative Events Scale–Revised

Author Note

1

Department of Psychology, Miami University

2

AN US

School of Social Service Administration, The University of Chicago

CR IP T

Lauren N. Forrest,1* Elizabeth A. Velkoff, 1 Courtney J. Johnson,2 Aaron Luebbe, 1 & April R. Smith1

*Correspondence concerning this article should be addressed to Lauren Forrest, Department of Psychology, Miami University, Oxford, OH, 45056, USA. Telephone: +1 513-529-2400; Fax: +1 513-

M

529-2420; Email: [email protected]

ED

Abstract

Background: Engagement in painful and provocative events is central to hypotheses for how capability

PT

for suicide develops. However, the existing measure of painful and provocative events is not psychometrically sound. We developed a measure with improved psychometric properties: the Painful

CE

and Provocative Events Scale (PPES)–Revised. Method: In Study 1, 447 adults (53.5% women, mean age = 35.4 years) answered 77 items describing painful and/or provocative experiences. Exploratory

AC

factor analysis (EFA) was performed. In Study 2, 403 adults (55.1% women, mean age = 38.1 years) answered the retained items and confirmatory factor analysis (CFA) was performed. The scale’s factorial

invariance across gender was examined. Estimates of convergent and discriminant validity were obtained. Results: EFA yielded a two-factor structure, which was confirmed with CFA. The factor structure was invariant across men and women. Estimates of convergent and discriminant validity were promising. Limitations: Limitations include non-weighted items, additional need for CFA among high-risk groups,

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

3

lack of assessment of person-specific painful and provocative events, and need for prospective research to establish the scale’s predictive validity. Conclusions: By assessing painful and provocative events more uniformly and reliably than the existing measure, the PPES–Revised has the potential to advance the

CR IP T

understanding of capability for suicide.

Keywords: painful and provocative events; capability for suicide; Interpersonal–Psychological Theory of

AC

CE

PT

ED

M

AN US

Suicide

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

4

Globally, nearly one million people die by suicide each year (World Health Organization, 2015). Despite tremendous advances in suicidology in recent decades, suicide rates have not declined and prediction of suicide attempts and deaths is poor (Franklin et al., 2017; Ribeiro et al., 2015). For

CR IP T

prediction to improve, theoretical models of suicidal behavior must be tested more comprehensively (Franklin et al., 2017). Modern suicide theories (e.g., Klonsky & May, 2015; Joiner, 2005; O’Connor, 2011; Van Orden, Cukrowicz, Witte, Braithwaite, Selby, & Joiner, 2010) propose that for people to die by suicide, individuals must desire suicide and have the capability to make a lethal attempt (hereafter

AN US

referred to as capability for suicide). Capability for suicide appears to be capable of upward or downward fluctuations occurring over brief or extended time periods (Zuromski, Cero, & Witte, 2018), and is thought to fluctuate in part through repeated painful and provocative experiences (e.g., risky and impulsive behaviors). Painful and provocative events are typically assessed with the Painful and

M

Provocative Events Scale (PPES; Van Orden, Witte, Gordon, Bender, & Joiner, 2008). However, multiple studies indicate that the PPES has an inconsistent and non-replicable factor structure (Brown, Roush,

ED

Marshall, Mitchell, & Cukrowicz, 2018; Poindexter, Nazem, & Forster, 2017; Teismann, Forkmann, Wachtel, Edel, Nyhuis, & Glaesmer, 2015; described further below). Given that painful and provocative

PT

events are a critical component of suicide theories, painful and provocative events must be measured accurately to test theoretically-informed hypotheses about suicide. Thus, the current study aimed to

CE

develop a psychometrically sound measure of painful and provocative events. According to the Interpersonal–Psychological Theory of Suicide (IPTS; Joiner, 2005; Van Orden

AC

et al., 2010), capability for suicide develops in part through repeatedly engaging in experiences that are painful and provocative, such as engaging in nonsuicidal self-injury (NSSI) or shooting a gun. Over time, repeated exposure to painful and provocative events is thought to increase pain tolerance and lower fear of death. Together, pain tolerance and fearlessness about death comprise capability for suicide. The experiences that lead to capability for suicide (i.e., painful and provocative events) are thought to be distinct from the processes that lead to suicidal ideation. For example, someone may have experienced

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

5

multiple painful injuries and thus have elevated capability for suicide; however, they may not have any suicidal ideation and therefore will be at low risk for suicide. However, when capability for suicide is experienced concurrently with active suicidal ideation, people are at elevated risk for highly lethal suicide attempts. In other words, painful and provocative events are thought to play a causal role in developing

CR IP T

capability for suicide, and capability for suicide—when experienced with suicidal ideation—is necessary to engage in suicidal behavior.

Painful and provocative events are often assessed using the PPES (Van Orden et al., 2008). The PPES is a self-report measure where respondents indicate on a Likert-type scale that ranges from 0

AN US

(never) to 4 (more than 20 times) how many times they engaged in different painful or provocative

experiences in their lifetimes, such as going rock climbing, tying a noose, or shooting a gun. The PPES was developed to advance suicide research, by measuring experiences theorized to play a key role in developing capability for suicide. However, this measure has significant limitations. First, different

M

versions of the PPES are often used to quantify painful and provocative experiences. As Poindexter and colleagues (2017) review, researchers have used versions of the PPES that include anywhere from 10 to

ED

74 items. Moreover, many articles do not report which painful and provocative events are included in the PPES version used in the particular study, which results in inconsistent operationalization of the painful

PT

and provocative events construct.

Second, the PPES was not validated psychometrically (e.g., with exploratory factor analysis

CE

[EFA] and/or confirmatory factor analysis [CFA]) prior to its publication. Since the PPES’ publication, three factor analyses have been conducted (Brown et al., 2018; Poindexter et al., 2017; Teismann et al.,

AC

2015). Each EFA resulted in retaining different numbers of PPES items and factors, and the authors offered different conceptualizations of the retained factors. Moreover, when the EFAs were followed up with CFAs, the factor structure failed to replicate (Brown et al., 2018; Teismann et al., 2015). The lack of consistent factor structure and the non-replication between EFA and CFA indicates that a measure with stronger psychometric properties is needed to reliably assess painful and provocative events.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

6

Brown and colleagues (2018) suggest that one possible reason for the PPES’ poor psychometric properties is inaccurate factor conceptualization. All existing PPES psychometric studies model painful and provocative events as a reflective latent factor, where the latent factor is thought to cause engagement in painful and provocative events (e.g., there is an underlying construct that predisposes people to engage

CR IP T

in painful and provocative behaviors). In contrast, Brown and colleagues (2018) propose that painful and provocative events should be conceptualized as a formative latent factor. A formative conceptualization of painful and provocative events proposes that engagement in painful and provocative events causes the development of the latent painful and provocative events factor. In other words, in reflective models,

AN US

indicators (e.g., items on the PPES) are modeled as effects of the latent factor, whereas in formative models indicators are modeled as causes of the latent factor (Bollen & Bauldry, 2011). When the IPTS was originally proposed, capability for suicide was thought to be acquired

through repeated exposure to painful and provocative events; the theory allowed for genetic influences on

M

capability for suicide but did not emphasize them directly (Joiner, 2005; Van Orden et al., 2010). If capability for suicide was primarily acquired, reconceptualizing painful and provocative events as a

ED

formative latent factor would seem fitting. However, in the years since the IPTS was proposed, theoretical and empirical work has questioned whether and how capability for suicide is acquired. If capability for

PT

suicide is not primarily acquired, then reconceptualizing painful and provocative events as a formative latent factor (Brown et al., 2018) seems ill advised. For example, in the three-step theory, Klonsky & May

CE

(2015) emphasize that capability for suicide is acquired in part but is also influenced by practical (i.e., knowledge and access to lethal means) and dispositional (e.g., genetics) factors. Similarly, Smith and

AC

Cukrowicz (2010) emphasize that biological and genetic factors drive engagement in painful and provocative events, which then manifest as capability for suicide (i.e., decreased fear of death and increased pain tolerance). Indeed, Smith and colleagues (2012) examined the magnitude of genetic and environmental effects on capability for suicide. In their study, capability for suicide was operationalized from several items that could be considered painful or provocative events (e.g., “riding a bicycle recklessly,” “using any kind of weapon in a fight,” and thrill-seeking and adventurous behaviors [p. 62]).

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

7

Results revealed that the best-fitting model included only genetic and non-shared environmental effects, indicating that engagement in painful and provocative events and capability for suicide are genetically mediated. The theoretical emphasis on, and empirical evidence for, genetic contributions to painful and provocative events and capability for suicide imply that genetics predispose or cause people to engage in

CR IP T

painful and provocative events, which then increases pain tolerance and decreases fear of death. In fact, Chu and colleagues (2017) recommend that the construct of capability for suicide undergo a “conceptual and nomenclature shift” (p. 1315) by dropping the term acquired, to reflect the construct’s possible

genetic contributions. In sum, although Brown and colleagues suggested a formative conceptualization of

AN US

painful and provocative events due to capability for suicide being acquired, conceptualizing painful and provocative events and capability for suicide as a reflective factor with effect indicators is most in line with the evolving understanding of suicide capability.

Accordingly, we set about developing a revised, psychometrically-sound version of the PPES, in

M

which we conceptualize painful and provocative events as a reflective factor. To accomplish this aim we first generated a large pool of possible painful and provocative events, including the painful and

ED

provocative events assessed in the original PPES. Once the final item pool was identified, in Study 1 we conducted EFA to identify the factor structure of the revised PPES, hereafter referred to as the PPES–

PT

Revised. In Study 2 we conducted CFA to test whether the EFA-identified PPES–Revised factor structure could be replicated and whether the factor structure was invariant across gender. In addition, we obtained

CE

estimates of convergent and discriminant validity in both study samples. Study 1: Item Generation, Item Selection, and EFA

AC

Method

Participants and procedures. Participants were recruited on Amazon’s Mechanical Turk

(MTurk; N=492). Inclusion criteria were being 18 years of age or older and living in the United States. MTurk is currently the dominant platform for crowdsourcing research participants and a reliable resource for collecting high-quality data (Shapiro, Chandler, & Mueller, 2013; Chandler & Shapiro, 2016). To verify attentive survey completion, participants completed three attention checks. Participants who failed

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

8

two or more attention checks were excluded from the sample (n=45), resulting in a total sample size of 447. The majority of participants identified as women (53.5%), Caucasian (82.8%), and Non-Hispanic or Latino (92.4%). Mean age was 35.37 years. See Table 1 for full demographic characteristics. All participants provided informed consent prior to beginning the study. Participants then

CR IP T

completed surveys through a secure online system and were then debriefed. Participants were compensated for participation. Mean survey completion time was 35.37 minutes and median survey completion time was 18.4 minutes. All procedures were approved by the university’s Institutional Review Board.

AN US

Measures.

Item generation and selection. Item generation for the PPES–Revised followed best-practice recommendations (Clark & Watson, 1995), by reviewing relevant literature, generating a large and overinclusive item pool, refining the item pool, and subjecting the refined item pool to factor analysis.

M

Specifically, after reviewing the literature on painful and provocative events and capability for suicide, several of the authors (L.N.F., E.A.V., A.R.S.) and one graduate research assistant generated a pool of

ED

162 painful and/or provocative experiences and behaviors, that were conceptually related to capability for suicide or suicide correlates. The items fell into eight categories, several of which overlapped with

PT

categories identified in Smith and colleague’s (2016) conceptual sorting of painful and provocative events: lifestyle; risky, provocative, or sensation-seeking; self-inflicted; other-inflicted; causing or

CE

witnessing others’ pain, suicide attempt, or death; disordered eating; and physical symptoms, diseases, or medical procedures. Additional information about item generation can be found in the Supplemental

AC

Materials. Included in the 162 items were the 26 items from the original PPES. All items were scored using the same Likert-type scale as the PPES, where respondents indicate how many times in their lifetime they experienced each event. Answer options are 0 (never), 1 (once), 2 (2–3 times), 3 (4–20 times), or 4 (more than 20 times). Item generation for scale development should be an iterative process (Clark & Watson, 1995). Thus, after generating items, we removed items that were duplicates, normative (e.g., exercising >30

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

9

minutes/day), too broad to be related to suicide (e.g., when you were a child, were you ever separated from your parents or guardians while in public), too rare for most people to provide a non-zero response (e.g., have you been in a plane crash), or contingent on another characteristic (e.g., have you had painful menstrual cramps [contingent on having menstrual cramps and being female]). Contingent items were

CR IP T

excluded to increase the scale’s generalizability. The refined item pool that was used in EFA consisted of 77 items, which included 22 of the original PPES items. All items included in the EFA, along with the frequencies and percentages of response options, are provided in Supplemental Table 1.

Validity. Because we did not have hypotheses about the factors that would be identified in the

AN US

EFA, we included measures related to the IPTS constructs (i.e., capability for suicide [assessed via selfreported pain tolerance and fearlessness about death], sensation seeking, perceived burdensomeness, thwarted belongingness, and suicidal ideation) to obtain preliminary estimates of convergent and/or discriminant validity for the factors that emerged.

M

Pain tolerance. Self-reported pain tolerance was assessed with the Pain Tolerance Evaluation Questionnaire (Rokke, Fleming-Ficek, Simens, & Hegstand, 2004). The questionnaire describes 12

ED

painful events and participants indicate whether they could tolerate this pain better than a given percentage of people of the same age and sex (e.g., I can tolerate a broken bone better than ____ percent

PT

of people who are the same age and sex). Answers for all items were averaged to create a single pain tolerance score, ranging from 0 to 100. Higher scores indicated higher self-reported pain tolerance.

CE

Reliability was excellent (α=.92).

Fearlessness about death. Fearlessness about death was assessed with the Acquired Capability

AC

for Suicide Scale–Fearlessness about Death subscale (Ribeiro, Witte, Van Orden, Selby, Gordon, Bender, & Joiner, 2014; Van Orden et al., 2008). The Acquired Capability for Suicide Scale–Fearlessness about Death subscale presents seven descriptions of having low fear of dying (e.g., I am not disturbed by death being the end of life as I know it). Participants indicated how true each statement was for them on a fiveitem Likert-type scale, with answers ranging from 0 (not at all like me) to 4 (very much like me). Items

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

10

were summed to create a total fearlessness about death score. Higher scores indicated greater fearlessness about death. Reliability was good (α=.89). Sensation seeking. Sensation seeking was assessed with the UPPS–Sensation Seeking subscale (Whiteside & Lynam, 2001). This self-report scale presents 12 statements that describe sensation-seeking

CR IP T

behaviors (e.g., I quite enjoy taking risks and I would enjoy water skiing). Participants indicate on a sixitem Likert-type scale how true each statement was for them, with answers ranging from 0 (not true of me) to 5 (very true of me). Items were summed to a total sensation-seeking score. Higher scores indicated greater sensation seeking. Reliability was excellent (α=.93).

AN US

Perceived burdensomeness and thwarted belongingness. Perceived burdensomeness and thwarted belongingness were assessed with the Interpersonal Needs Questionnaire (Van Orden, Cukrowicz, Witte & Joiner, 2012). This self-report scale includes six items assessing perceived liability and nine items assessing low belongingness. Participants indicated on a seven-item Likert-type scale how true each

M

description was for them currently, with answers ranging from 0 (not at all true for me) to 6 (very true for me). Relevant items were summed to create a total score for perceived burdensomeness and a total score

ED

for thwarted belongingness. Higher scores indicated greater perceived burdensomeness and thwarted belongingness. Reliability was excellent for perceived burdensomeness (α=.96) and thwarted

PT

belongingness (α=.94).

Suicidal ideation. The Depressive Symptom Severity Index–Suicidality Subscale assessed the

CE

level of suicide ideation participants experienced over the previous two weeks (Joiner, Pfaff, & Acres, 2002). The four-item, self-report scale assesses the controllability of suicidal ideation, the frequency of

AC

suicidal ideation, planning a suicide attempt, and the intensity of the impulses to attempt suicide. Each item includes four descriptions of suicidal ideation severity that are scored from 0 to 3, and participants select the description that was most true for them over the previous two weeks. Higher scores indicate greater suicidal ideation severity. Reliability was excellent (α=.92). Data analyses. We first inspected whether data were appropriate for EFA by examining the Kaiser-Meyer-Olkin index, which determines the sampling adequacy. EFA was then performed in Mplus

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

11

version 7 (Muthén & Muthén, 1998–2012). Weighted least squares with mean and variance adjustment (WLSMV) was used as the estimator, given that some items were expected to be distributed non-normally (e.g., attempting suicide). In addition, WLSMV is the recommended approach for Likert-type data (Li, 2016). Geomin rotation was performed, specifying up to 10 factors. The number of factors to retain was

CR IP T

determined by examining the scree plot and inspecting model fit statistics. The scree plot was used to identify the number of factors above the “elbow,” by visually inspecting plotted eigenvalues. Model fit statistics included model χ2, Root Mean Square Error of Approximation (RMSEA), Comparative Fit Index (CFI), and Tucker Lewis Index (TLI). Model chi-square values with ps<.05 were not automatically

AN US

rejected if other fit statistics indicated good or excellent fit. Based on previous recommendations, “good” fit was indicated by RMSEA<.08, CFI≥.90, and TFI≥.90, while “excellent” fit was indicated by RMSEA<.05, CFI>.95, and TLI>.95 (Hu & Bentler, 1999).

After the best fitting model was identified, individual items were retained if they loaded ≥.32 on

M

the relevant factor but <|.32| on all other factors, given that we wanted to achieve simple structure (i.e., identify a model where each item loads significantly onto only one factor; Tabachnick & Fidell, 2001).

ED

After identifying items to be retained, the number of remaining indicators per factor was examined. Factors were retained if they had three or more indicators, given that two-item factors are not identified

PT

on their own. This process was repeated until simple structure was achieved. After achieving simple structure, the inter-item correlations were inspected. Positive inter-item correlations are consistent with

CE

the reflective construct operationalization (Bollen & Bauldry, 2011). Missing data were minimal and missingness was therefore handled with the pairwise present

AC

option in Mplus.

We obtained initial estimates of convergent and discriminant validity by computing correlations

among the PPES–Revised factor totals and pain tolerance, fearlessness about death, sensation seeking, perceived burdensomeness, thwarted belongingness, and suicidal ideation. We expected that the PPES– Revised factors would demonstrate medium, positive correlations with pain tolerance, fearlessness about death, and sensation seeking. Given that capability for suicide can develop independently from suicidal

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

12

ideation, we expected either nonsignificant or small, positive correlations between PPES–Revised factors and perceived burdensomeness, thwarted belongingness, and suicidal ideation. Results The Kaiser-Meyer-Olkin index was 0.84, indicating adequate sampling accuracy to conduct EFA

CR IP T

(Dziuban & Shirkey, 1974). The scree plot indicated a five- or six-factor solution. Eigenvalues for the retained factors in the five- and six-factor solutions were ≥2. Fit statistics were good for each solution (five-factor: model χ2[2551]=3306.21, p<.001; RMSEA=.03, 90% CI [.02, .03]; CFI=.95; TLI=.94; sixfactor: model χ2[2479]=3035.44, p<.001; RMSEA=.02, 90% CI [.02, .03]; CFI=.96; TLI=.95). Given the

AN US

similarity in fit statistics, the five-factor solution was retained. Retaining the five-factor solution resulted in dropping 51 items due to item factor loadings <.32 or crossloadings ≥ |.32|. After dropping the 51 items, the final scree plot indicated retaining two factors. Eigenvalues for the two retained factors were ≥3.5 and fit was good-to-excellent (model χ2[274]=505.86, p<.001; RMSEA=.04, 90% CI [.04, .05];

M

CFI=.94; TLI=.93). Geomin-rotated factor loadings for the retained 26 items are presented in Table 2. Only six of the 26 original PPES items were retained.

ED

The first factor contained 17 items that assessed experiences that resulted in injury or had the potential to result in injury (=.76). Factor 1 was thus labeled as Actual or Potential Injuries. The second

PT

factor contained nine items that assessed self-harming experiences, including direct self-harm (e.g., deliberately harming oneself), indirect self-harm (e.g., self-induced vomiting and fasting), and thinking

CE

about self-injury (=.78). Factor 2 was thus labeled as Self-harm. Internal consistency for all PPES– Revised items was good (=.81). All inter-item correlations within each factor were positive

AC

(Supplemental Tables 2 and 3). The 26-item scale and scoring instructions are provided in the Supplemental Material. Items were summed to create a total score for each factor. The correlation between factor totals

was positive, moderate, and statistically significant (Table 3). This suggests that painful and provocative experiences are best conceptualized as a multi-factor construct.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

13

Preliminary estimates of convergent and discriminant validity. Correlations were computed to establish preliminary convergent and discriminant validity of the two-factor PPES–Revised (Table 3). The Actual or Potential Injuries factor exhibited small to moderate positive correlations with pain tolerance, fearlessness about death, sensation seeking, and suicidal ideation. The Actual or Potential

CR IP T

Injuries factor did not significantly correlate with perceived burdensomeness or thwarted belongingness. The Self-harm factor exhibited small positive correlations with pain tolerance and fearlessness about death, and moderate positive correlations with perceived burdensomeness, thwarted belongingness, and suicidal ideation. The Self-harm factor did not significantly correlate with sensation seeking.

AN US

Study 2: CFA

In a second sample, we administered the PPES–Revised to accomplish three aims. First, we aimed to confirm the two-factor structure identified in Study 1, using CFA. We predicted that the factor structure would replicate and would provide good fit to the data.

M

Second, we examined factorial invariance of the PPES–Revised between men and women, because the original PPES has been used across genders (e.g., Teismann et al., 2015; Zuromski et al.,

ED

2018) and we must determine whether gender differences exist in factor structure and factor means. That is, to ensure the scale’s appropriateness to be used for men and women, we would expect for men and

PT

women to display similar PPES–Revised factor structures. However, we may expect for gender differences to emerge in factor means, given that men exhibit greater engagement in sensation seeking

CE

(Witte, Gordon, Smith, & Van Orden, 2012) and have higher capability for suicide than women (Ribeiro et al., 2014). Accordingly, we predicted that men would have a higher mean on the Actual or Potential

AC

Injuries factor relative to women. We did not predict that men and women would display mean differences on the Self-harm factor. While three items on the Self-harm factor assess disordered eating, which are more commonly experienced among women versus men, other items were expected to display no gender differences (e.g., engaging in NSSI; Klonsky, 2011), while others still are more commonly experienced among men versus women (e.g., experiencing violent daydreams and fantasies). Given that

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

14

the factor mean reflects all items, we expected that the mixed gender differences on individual items would not be reflected in the overall Self-harm factor mean. Third, we aimed to establish the convergent and discriminant validity of the PPES–Revised. Notably, while we included preliminary estimates of convergent and discriminant validity in Study 1, in

CR IP T

Study 2 we selected additional and specific indicators of validity based on the factors that emerged in the EFA. To establish convergent validity, we predicted that both factors would correlate positively with components of capability for suicide (i.e., pain tolerance and fearlessness about death), sensation seeking, the number (i.e., variety) of methods used to engage in NSSI, disordered eating, and traumatic life events.

AN US

We also assessed how the factors correlated with perceived burdensomeness, thwarted belongingness, and suicidal ideation. We predicted that these constructs would correlate positively with the Self-harm factor (i.e., serve as additional indicators of convergent validity) yet negatively with the Actual or Potential Injuries factor (i.e., serve as indicators of discriminant validity). Please see the Supplementary Materials

M

for a detailed justification for including each convergent and discriminant validity indicator. Method

ED

Participants and procedures. Participants for Study 2 were recruited on MTurk (N=428). Inclusion criteria were being 18 years of age or older and living in the United States. As in Study 1,

PT

participants completed three attention checks to ensure high-quality data. Participants who failed two or more attention checks (n=20), had duplicate IP addresses (n =3), or had an IP address that was also

CE

recorded in the Study 1 sample (n=2) were excluded, resulting in a final sample size of 403. The majority of participants identified as women (55.1%), Caucasian (81.1%), and Non-Hispanic or Latino (91.1%).

AC

Mean age was 38.08 years. See Table 1 for full demographic characteristics. All participants provided informed consent prior to beginning the study. Upon providing consent,

participants completed surveys through a secure online system and were then debriefed. Participants were compensated for participation. Mean survey completion time was 16.3 minutes and median survey completion time was 14.1 minutes. All procedures were approved by the university’s Institutional Review Board.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

15

Measures. PPES–revised. The 26 items assessing painful and provocative events that were retained in the Study 1 EFA (i.e., the PPES–Revised) were administered. As described above, respondents indicated how many times they experienced each event in their lifetime. Reliability was good for the Actual or Potential

CR IP T

Injuries factor (α=.87) and acceptable for the Self-harm factor (α=.79). Validity. Several Study 1 measures used to establish convergent and discriminant validity were also used in Study 2. Please see scale descriptions provided in Study 1.

Pain tolerance. Self-reported pain tolerance was assessed with the pain tolerance item (I can

indicate greater self-reported pain tolerance.

AN US

tolerate a lot more pain than most people) on the Acquired Capability for Suicide Scale. Higher values

Fearlessness about death. Fearlessness about death was assessed with the Acquired Capability for Suicide Scale–Fearlessness about Death subscale (Ribeiro et al., 2014; Van Orden et al., 2008).

M

Reliability was good (α=.88).

Sensation seeking. Sensation seeking was assessed with the UPPS–Sensation Seeking subscale

ED

(Whiteside & Lynam, 2001). Reliability was excellent (α=.94). NSSI. The methods used to engage in NSSI in the past year were assessed with the Functional

PT

Assessment of Self-Mutilation (Lloyd, Kelley, & Hope, 1997). The Functional Assessment of SelfMutilation presents 12 NSSI methods (e.g., cutting skin, picking at wound) and respondents indicate

CE

whether they engaged in the behaviors in the past year. The number of NSSI methods used in the past year was summed. Reliability was acceptable (α=.73).

AC

Disordered eating. Disordered eating was assessed with the SCOFF (Morgan, Reid, & Lacey,

1999). This self-report scale is often used to screen for eating disorder risk. Thus, it includes five descriptions of heterogeneous disordered eating symptoms (e.g., do you make yourself sick because you feel uncomfortably full? and have you recently lost more than one stone [14 lb] in a 3-month period?). Participants indicate whether they recently experienced these symptoms (i.e., answers are rated as yes or no). SCOFF items were summed, and higher total scores indicate experiencing more disordered eating

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

16

symptoms. Reliability was questionable (α=.64). However, low internal consistency is expected because the SCOFF assesses distinct symptoms that are not always experienced concurrently (e.g., recent weight loss is not necessarily expected to be correlated with experiencing loss of control over eating) and is not designed to be a homogenous scale.

CR IP T

Traumatic life experiences. Traumatic life experiences were assessed with the Life Events Checklist (Gray, Litz, Hsu, & Lombardo, 2004). This self-report scale presents 17 descriptions of

potentially traumatic life events (e.g., being in a fire, witnessing death). Participants indicate whether they experienced each event in their lifetime. Traumatic life experiences were summed. Reliability was

AN US

acceptable (α=.79).

Perceived burdensomeness and thwarted belongingness. Perceived burdensomeness and thwarted belongingness were assessed with the Interpersonal Needs Questionnaire (Van Orden et al., 2012). Reliability was excellent for perceived burdensomeness (α=.97) and thwarted belongingness (α=.90).

M

Suicidal ideation. The Depressive Symptom Severity Index–Suicidality Subscale assessed the level of suicide ideation participants experienced over the previous two weeks (Joiner et al., 2002).

Data analytic plan.

ED

Reliability was excellent (α=.92).

PT

Aim 1: CFA model fitting. We first inspected whether data were appropriate for CFA by examining the Kaiser-Meyer-Olkin index. CFA was then performed in Mplus version 7 (Muthén &

CE

Muthén, 1998–2012). As in Study 1, WLSMV was used as the estimator. Fixed factor scaling was used, such that a parameter estimate was generated for each factor indicator. Also as in Study 1, model fit was

AC

evaluated with model χ2, RMSEA, CFI, and TLI, using the same criteria to indicate good vs. excellent fit. No data were missing. Aim 2: Factorial invariance. Three levels of factorial invariance between men and women were

examined using multiple group analysis with theta parameterization. The three levels included configural, metric (i.e., “weak”) and scalar (i.e., “strong”) invariance (Brown, 2014). Testing for each level of invariance proceeded sequentially by applying parameter constraints to factor variances, factor means,

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

17

factor loadings, and item thresholds (i.e., the number of response categories in each item). At each level, residual variances were constrained to 1 across groups, as constraining them to equality between groups can result in an overly strict invariance criterion (Little & Slegers, 2005; Byrne, 2012). We computed the χ2 difference test using the DIFFTEST option in Mplus between each level of invariance. However,

CR IP T

because the χ2 difference test is overly strict, invariance outcomes were not based solely on these results (Chen, 2007). Specifically, even if the χ2 difference test was significant, metric invariance was indicated if the CFI reduction was < .01 and if RMSEA increased < .015 relative to the configural model (Chen, 2007). Similarly, even if the χ2 difference test was significant, scalar invariance was indicated if the CFI

AN US

reduction was < .01 and if RMSEA increased < .015 relative to the metric model (Chen, 2007).

Aim 3: Convergent and discriminant validity. We established convergent and discriminant validity by computing correlations among the PPES–Revised factor totals and pain tolerance, fearlessness about death, sensation seeking, NSSI methods, disordered eating, traumatic life experiences, perceived

M

burdensomeness, thwarted belongingness, and suicidal ideation. Results

ED

Aim 1: CFA model fitting. The Kaiser-Meyer-Olkin index was 0.91, indicating excellent sampling accuracy to conduct CFA (Dziuban & Shirkey, 1974). The frequencies and proportions for all

PT

response options for each item are shown in Supplemental Table 4. The two-factor structure provided good fit to the data (Table 4). All items loaded significantly onto their assigned factor (ps<.001; see

CE

Figure 1). Internal consistency for all 26 PPES–Revised items was excellent (=.90). Three modification indices were suggested but not included because model fit was good prior to specifying modification

AC

indices and because modification indices capitalize on chance in a sample and may not replicate across samples.

As in Study 1, all inter-item correlations within each factor were positive (Supplemental Tables 2

and 3). The two PPES–Revised factors correlated positively and significantly (Table 5). However, unlike in Study 1, the correlation between factors was large in magnitude. In sum, the EFA-specified model was replicated in the CFA sample.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

18

Aim 2: Factorial invariance. Configural invariance. Testing for configural invariance identifies whether groups have the same factor structure. In this model, the same factor structure was estimated simultaneously in men and women. In both groups, factor variances were constrained to 1 while factor means were constrained to 0, to allow

CR IP T

for factor loadings and item thresholds (i.e., the number of response categories in each item) to be estimated freely in each group. Residual variances were constrained to 1 across groups. The configural invariance model provided good-to-excellent fit (χ2 contribution for women=680.81, χ2 contribution for men=503.44; Table 4).

AN US

Metric factorial invariance. Testing for metric invariance identifies whether factor loadings differ between groups. In this model, factor variances were constrained to 1 in women but estimated freely in men. In both groups, factor means were constrained to 0 and factor loadings were constrained to equality. Item thresholds were estimated freely in both groups. Residual variances remained constrained

M

to 1. The χ2 difference test was significant but the change in CFI was <.01 and the change in RMSEA was <.015, indicating that the PPES–Revised items were estimated equivalently among men and women.

ED

Scalar factorial invariance. Testing for scalar factorial invariance identifies whether item thresholds and/or means differ across groups. In this model, factor variances were constrained to 1 and

PT

factor means were constrained to 0 in women, but factor variances and means were estimated freely among men. All factor loadings and item thresholds were constrained to equality across groups, and

CE

residual variances remained constrained to 1 in both groups. The χ2 difference test was significant but the change in CFI was <.01 and the change in RMSEA was <.015 (Table 4), indicating that item thresholds

AC

were similar across groups. However, factor means inspection revealed that, relative to women, men had a descriptively higher mean for the Actual or Potential Injuries factor (p =.09). Aim 3: Convergent and discriminant validity. Convergent validity. Correlations were inspected to establish convergent validity of the PPES– Revised (Table 5). In support of hypotheses, the Actual or Potential Injuries factor exhibited small positive correlations with pain tolerance, fearlessness about death, sensation seeking, NSSI methods, and

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

19

disordered eating, and a large positive correlation with traumatic life events. The Self-harm factor exhibited small positive correlations with pain tolerance, sensation seeking, and thwarted belongingness, and moderate positive correlations with NSSI methods, disordered eating, traumatic life events, perceived burdensomeness, and suicidal ideation. However, contrary to hypotheses, the Self-harm factor had a very

CR IP T

small and nonsignificant correlation with fearlessness about death. Discriminant validity. Correlations were inspected to establish discriminant validity for the Actual or Potential Injuries factor. In partial support of expectations, the Actual or Potential Injuries factor exhibited a small negative correlation with thwarted belongingness but small positive correlations

AN US

with perceived burdensomeness and suicidal ideation. General Discussion

We aimed to develop an improved assessment of painful and provocative events, given that the existing measure shows poor psychometric properties. After generating over 150 novel painful and

M

provocative events, in Study 1 we conducted EFA to identify the factor structure of the PPES–Revised. In Study 2 we conducted CFA to confirm the EFA-identified PPES–Revised factor structure. The identified

ED

factors assess forms of painful and provocative events that are related to risk factors for suicidal behavior. Moreover, the measure appears to have an invariant factor structure across men and women, and

PT

estimates of convergent and discriminant validity for the factors are encouraging. Below we describe literature supporting links between the PPES–Revised factors with other constructs related to capability

CE

for suicide and suicidal behavior, discuss the PPES–Revised factorial invariance, and provide suggestions for future studies to further test and refine the PPES–Revised.

AC

PPES–Revised Factors and Convergent and Discriminant Validity In Study 1, we conducted EFA with 77 items assessing different forms of painful and provocative

events. Results indicated that a 26-item structure with two factors provided good fit to the data. The identified factors were Actual or Potential Injuries and Self-harm. The factor structure was replicated in the Study 2 CFA. The Actual or Potential Injuries factor included multiple items that may have been the result of sensation-seeking behavior, such as breaking a bone (e.g., due to playing an extreme or contact

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

20

sport) or being in a car accident (e.g., due to reckless driving). Sensation seeking is a personality trait associated with engaging in risky or fearsome behaviors, or engaging in activities that may result in injury or death, such as riding a motorcycle or shooting a gun (e.g., Zuckerman, Eysenck, & Eysenck, 1978). Sensation seeking is closely associated with the stereotypical male gender role, and men often display

CR IP T

greater sensation seeking than women (e.g., Zuckerman et al., 1978). Sensation seeking is positively associated with capability for suicide (Anestis, Bagge, Tull, & Joiner, 2011; Bender, Gordon, Bresin, & Joiner, 2011; Witte et al., 2012), and meditational studies indicate that sensation seeking mediates the association between male gender and fearlessness about death (Witte et al., 2012).

AN US

Several items that loaded onto the Actual or Potential Injuries factor (e.g., seeing someone die or someone close to you dying by suicide) are not directly related to sensation-seeking behavior. However, seeing someone die or losing a loved one to suicide arguably inure individuals to death and likely provoke painful emotions (e.g., fear, sadness). Interestingly, even though these experiences may not involve

M

physical pain, neurobiological studies indicate that physical and emotional pain are processed using the same substrates in the brain (for a review, please see Eisenberger, 2012). Thus, the inurement to death and

ED

the emotional pain involved in witnessing death or losing a close individual to suicide appear to be consistent with the IPTS’ predictions that painful and provocative experiences habituate individuals to

PT

fear of death and pain (Joiner, 2005; Van Orden et al., 2010). Of note, the Actual or Potential Injuries factor was the only factor for which both convergent and

CE

discriminant validity could be estimated. The pattern of results was similar in the Study 1 and Study 2 samples, where the factor correlated positively and significantly with fearlessness about death, self-

AC

reported pain tolerance, and sensation seeking (i.e., all indicators of convergent validity). Further, in Study 2 the Actual or Potential Injuries factor correlated significantly and positively with additional indicators of convergent validity: NSSI methods, disordered eating, and traumatic life events. In partial support of expectations, the Actual or Potential Injuries did not correlate significantly with thwarted belongingness in either study, yet correlated positively and significantly with perceived burdensomeness in Study 2 and with suicidal ideation in both studies. Although not required theoretically by the IPTS,

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

21

several other studies have found significant relations between capability for suicide and suicidal ideation (e.g., George, Page, Hooke, & Stritzke, 2016). These findings may be due to cross-sectional designs, as these designs do not provide information about how or when the “routes” to suicidal behavior (desire and capability) manifested. If people have attempted suicide, we would expect for suicidal ideation to be

CR IP T

positively associated with capability for suicide and painful and provocative events because they are each associated with attempts. However, if people have not attempted suicide, have no suicidal ideation but have elevated capability, these constructs would not be associated. Ultimately, prospective research is needed to clarify how painful and provocative events and capability for suicide are related (or not) to

AN US

suicidal ideation.

The Self-harm factor included items that assess engaging in direct self-harm (e.g., aborted suicide attempts, NSSI), indirect self-harm (e.g., self-induced vomiting), and thinking about one’s death. Multiple studies indicate that each of these forms of self-injurious thoughts and behaviors are positively associated

M

with capability for suicide and/or risk for suicide attempts. For instance, previous self-injurious behavior (i.e., direct self-harm) is thought to be one of the most potent ways to increase capability for suicide and

ED

NSSI is one of the strongest predictors of future suicide attempts (Joiner, 2005; Ribeiro et al., 2015). Repeatedly engaging in disordered eating behaviors (i.e., indirect self-harm) is extremely painful, and is

PT

thought to increase capability for suicide among people with eating disorders (Joiner, 2005; see review in Witte et al., 2016). Indeed, individuals with eating disorders have elevated risk for suicide attempts and

CE

death (Chesney et al., 2014; Crow et al., 2009; Keshaviah et al., 2014). Mentally rehearsing dying by suicide is theorized as a possible provocative experience that increases capability for suicide (Joiner,

AC

2005). At least three studies indicate that mentally rehearsing painful and provocative events (e.g., reexperiencing traumatic events, experiencing violent daydreams, mentally rehearsing a suicide attempt) are positively associated with higher capability for suicide (Bryan & Anestis, 2011; George et al., 2016) and greater severity of suicidal ideation (Selby, Anestis, & Joiner, 2007). Thus, although the original PPES included only painful or provocative behaviors, the PPES–Revised reflects more recent work suggesting

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

22

that mental rehearsal of violent or death-related imagery can be considered provocative events capable of increasing capability for suicide. The relations among painful and provocative events, capability for suicide, and self-injurious behavior are central to the IPTS. The Self-harm factor correlated positively and significantly with pain

CR IP T

tolerance in both studies, yet correlated positively and significantly with fearlessness about death only in Study 1. Given this somewhat mixed correlation pattern, additional investigations are needed to further understand the relation between the Self-harm factor and facets of capability for suicide. However,

correlations with other constructs support the convergent validity of the Self-harm factor. In both studies,

AN US

the Self-harm factor correlated positively and significantly with perceived burdensomeness, thwarted belongingness, and suicidal ideation. In Study 1, the Self-harm factor also correlated positively and significantly with sensation seeking and in Study 2, the Self-harm factor correlated positively and significantly with NSSI methods, disordered eating, and traumatic life experiences.

M

Overall, many of the PPES–Revised items are conceptually consistent with three types of experiences—dispositional, practical, and acquired—which Klonsky and May (2015) propose are related

ED

to suicide capability. Dispositional factors are largely genetically driven, such as pain sensitivity or personality traits associated with suicide capability, like sensation seeking. As described above, sensation

PT

seeking is likely to drive engagement in several behaviors in the Actual or Potential Injuries factor. Practical factors are “concrete factors that make a suicide attempt easier…[f]or example, someone with

CE

both knowledge of and access to lethal means, such as a firearm, will be more able to act on suicidal thoughts than someone who lacks knowledge of and access to lethal means” (Klonsky & May, 2015, p.

AC

119). If someone has regularly thought about their own death in detail or collected items to prepare for serious self-harm (two items included on the PPES–Revised), it is reasonable to hypothesize that this individual will be more able to act on suicidal thoughts than someone who has never engaged in these behaviors (c.f. Hooley & Franklin, 2018). Finally, acquired factors are related to experiencing painful and provocative events, similar to what Joiner (2005) and Van Orden and colleagues (2010) proposed originally for what contributes to suicide capability. As described above, many of the items generated for

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

23

the PPES–Revised were identified because they were thought to involve some degree of pain and/or provocation. Taken together, to fully understand fluctuations in suicide capability, we must consider how dispositional, practical, and acquired factors interact. Because the PPES–Revised assesses experiences that fit within each of these categories, we believe that the measure has promise to meet this aim, though

CR IP T

whether or not the PPES–Revised is associated with changes in suicide capability is an empirical question to be addressed in future research. Factorial Invariance

In the Study 2 sample, we found that the PPES–Revised had an invariant factor structure across

AN US

gender. Specifically, the tests of metric factorial invariance indicated that the PPES–Revised items were estimated similarly for men and women. The fact that all items loaded positively and significantly onto their respective factors and explained a significant amount of variance for both men and women is notable. When testing for scalar invariance, all thresholds for all items were estimated similarly between

M

men and women. However, men had a descriptively higher latent mean for the Actual or Potential Injuries factor as compared to women. This gender difference is consistent with IPTS predictions and existing

ED

self-report measures of capability for suicide, where men have higher latent means for fearlessness about death relative to women (Ribeiro et al., 2014).

PT

Limitations and Future Directions

The study has several limitations. Two limitations relate to assessing the frequency of painful and

CE

provocative events. First, the frequency cutpoints employed here were not empirically identified, and a different response scale could impact the EFA and CFA results. Exploratory data mining techniques could

AC

be used to identify meaningful cutpoints for the frequencies of painful and provocative events that differentially affect capability for suicide. Second, frequencies do not capture the degree of pain and provocation experienced during events. Critically, painful and provocative events are thought to increase pain tolerance and decrease fear of death through opponent processes: over time and with repetition, the primary process must be elicited (e.g., fear of death) for the opposite process (e.g., fearlessness of death) to become amplified. Currently, only the repetition of events is assessed, yet the degree of pain, fear, and

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

24

provocation experienced during events may yield key information about how painful and provocative events affect capability for suicide and suicidal behavior (Smith & Cukrowicz, 2010). Third, painful and provocative events are thought to confer differential increases in capability for suicide, where events more directly related to suicidal behavior yield the most potent increases in

CR IP T

capability for suicide (Joiner, 2005; Smith & Cukrowicz, 2010). Weighting PPES–Revised items by their frequencies, degree of pain and provocation, and/or approximation of suicidal behavior may be useful. However, the current study included unweighted items, because reliable and valid item weights would need to be derived from multiple samples with varying degrees of painful and provocative events and

AN US

varying degrees of suicide risk. We included only two samples of unselected participants, with few

participants who reported lifetime suicidal behavior (9.6% in Study 1 [n = 43], 7.6% in Study 2 [n = 31]; lifetime suicide attempts were assessed in both studies using the following question: have you ever attempted suicide?). These samples were sufficient for conducting EFA and CFA but they are not well

M

matched to the goal of identifying meaningful item weights. Relatedly, the factor means in Study 1 were descriptively greater than the means in Study 2. Additional confirmatory studies are needed to identify

ED

average factor means and structures among norm and high-risk groups. Fourth, psychological science often seeks to maintain a balance between nomothetic and

PT

idiographic approaches. The PPES–Revised aimed to assess general painful and provocative events, yet person- or group-specific painful and provocative events (e.g., carrying a weapon as part of one’s job,

CE

performing euthanasia on animals [Witte, Correia, & Angarano, 2013]) may also be associated with fluctuations in capability for suicide and elevated risk for suicide attempts. Future research would benefit

AC

from addressing how both general and person- or group-specific painful and provocative events relate to capability for suicide. Relatedly, the current study included American participants who mainly identified as non-Hispanic and white. Findings (e.g., the events included in the PPES–Revised, the factor structure of the PPES–Revised, etc.) may not generalize to Americans with other racial and ethnic identities or to people from other countries.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

25

Fifth, even though the two metrics of self-reported pain tolerance was significantly and positively correlated with both factors in the studies, behavioral measures of pain tolerance may yield a more precise and ecologically valid operationalization of the construct (Chu et al., 2017). Future studies would benefit from examining relations between the PPES–Revised factors and behaviorally assessed pain tolerance.

CR IP T

Sixth, prospective research is needed to establish the predictive validity of the PPES–Revised. Ultimately, prospective research is the gold standard to address several of the limitations noted here: whether painful and provocative events are best indicated by the frequency of events and/or degree of pain and provocation elicited during events, whether and how items should be weighted, and whether

AN US

general or specific painful and provocative events are better predictors of capability for suicide and suicidal behavior.

Limitations notwithstanding, the project has notable strengths and implications for future suicide research. The development of the PPES–Revised adhered to best-practice recommendations for scale

M

development (Clark & Watson, 1995), including conducting EFA and CFA on separate samples, identifying factorial invariance across gender, and establishing estimates of convergent and discriminant

ED

validity. The studies presented here indicate that the PPES–Revised has the psychometric properties needed to reliably assess painful and provocative events, which the original version of the PPES lacks

PT

(Brown et al., 2018; Poindexter et al., 2017; Teismann et al., 2015). Given the use of EFA and CFA, we identified specific higher-order forms of painful and provocative events. Our identification of two

CE

replicable, higher-order factors will allow research to examine the potency of specific forms of painful and provocative experiences in increasing capability for suicide (Joiner, 2005). In addition, using the

AC

PPES–Revised rather than the original version will allow for consistent operationalization of painful and provocative events, given that the revised measure clearly delineates that 26 items are assessed, rather than a range of 10–74 items (Poindexter et al., 2017). Summary In sum, painful and provocative events are theorized to play a critical role in developing capability for suicide, which in turn is theorized to be necessary for engaging in suicidal behavior. To test

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

26

theoretical models of suicidal behavior—and to advance the prevention, prediction, and treatment of suicide—we must use measures with strong psychometric properties. The PPES–Revised demonstrates a replicable factor structure that is invariant across men and women. Further, it has promising estimates of convergent validity. By assessing painful and provocative events more uniformly and reliably than the

capability for suicide relates to suicidal behavior.

Author Statement

CR IP T

existing measure, the PPES–Revised has the potential to advance the field’s understanding of how

AN US

Author Contributions Lauren Forrest, Elizabeth Velkoff, Courtney Johnson, and April Smith developed the research questions. Elizabeth Velkoff and Courtney Johnson assisted with data collection. Lauren Forrest performed statistical analyses, and April Smith and Aaron Luebbe provided statistical consultation. Lauren Forrest drafted the manuscript. All study authors contributed to, edited, and reviewed the manuscript and approved of the manuscript in its present form.

M

Role of Funding Source This research did not receive any specific grant from funding agencies in the public, commercial, or notfor-profit sectors.

ED

Acknowledgements None

Conflict of Interest

AC

CE

PT

All authors declare that they have no conflicts of interest.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

27

References Anestis, M. S., Bagge, C. L., Tull, M. T., Joiner, T. E., 2011. Clarifying the role of emotion dysregulation in the interpersonal–psychological theory of suicidal behavior in an undergraduate sample. J Psychiatr Res. 45, 603–611. DOI: 10.1016/j.jpsychires.2010.10.013

CR IP T

Bender, T. W., Gordon, K. H., Bresin, K., Joiner, T. E., 2011. Impulsivity and suicidality: the mediating role of painful and provocative experiences. J Affect Disord. 129, 301–307. DOI: 10.1016/j.jad.2010.07.023

Brown, S. L., Roush, J. F., Marshall, A. J., Mitchell, S. M., Cukrowicz, K. C., 2018. A psychometric

AN US

investigation of the Painful and Provocative Events Scale: Moving forward. Arch Suicide Res. 22, 628–643. DOI: 10.1080/13811118.2017.1392914

Brown, T. A., 2014. Confirmatory factor analysis for applied research. New York: Guilford Publications. Bryan, C. J., Anestis, M., 2011. Reexperiencing symptoms and the interpersonal–psychological theory of

M

suicidal behavior among deployed service members evaluated for traumatic brain injury. J Clin Psychol. 67, 856–865. DOI: 10.1002/jclp.20808.

ED

Byrne, B. M., 2012. Structural equation modeling with Mplus: Basic concepts, applications, and programming. New York, NY: Routledge.

PT

Chandler, J., Shapiro, D., 2016. Conducting clinical research using crowdsourced convenience samples. Annu Rev Clin Psychol. 12, 53–81. DOI: 10.1146/annurev-clinpsy-021815-093623

CE

Chesney, E., Goodwin, G. M., & Fazel, S., 2014. Risks of all-cause and suicide mortality in mental disorders: a meta-review. World Psychiatry. 13, 153–160. DOI: 10.1002/wps.20128.

AC

Chu, C., Buchman-Schmitt, J. M., Stanley, I. H., Hom, M. A., Tucker, R. P., Hagan, C. R., Rogers, M. L., Podlogar, M. C., Chiurliza, B., Ringer, F. B., Michaels, M. S., Patrox, C. H. G., Joiner, T. E., 2017. The interpersonal theory of suicide: A systematic review and meta-analysis of a decade of cross-sectional research. Psychol Bull. 143, 1313–1345. DOI: 10.1037/bul0000123 Clark, L. A., Watson, D., 1995. Constructing validity: Basic issues in objective scale development. Psychol Assess. 7, 309–319.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

28

Costello, A. B., Osborne, J. W., 2005. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical Assess Res Eval. 10, 1–9. Crow, S. J., Peterson, C. B., Swanson, S. A., Raymond, N. C., Specker, S., Eckert, E. D., Mitchell, J. E., 2009. Increased mortality in bulimia nervosa and other eating disorders. Am J Psychiatry. 166,

CR IP T

1342–1346. Dziuban, C. D., & Shirkey, E. C., 1974. When is a correlation matrix appropriate for factor analysis? Psychol Bull. 81, 358–361.

Chen, F. F., 2007. Sensitivity of goodness of fit indexes to lack of measurement invariance. Struct Equ

AN US

Model. 14, 464–504. DOI: 10.1080/10705510701301834

Franklin, J. C., Ribeiro, J. D., Fox, K. R., Bentley, K. H., Kleiman, E. M., Huang, X., Musacchio, K. M., Jaroszewski, A. C., Chang, B. P., Nock, M. K., 2017. Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychol Bull. 143, 187–232. DOI:

M

10.1037/bul0000084

George, S. E., Page, A. C., Hooke, G. R., Stritzke, W. G., 2016. Multifacet assessment of capability for

ED

suicide: Development and prospective validation of the Acquired Capability with Rehearsal for Suicide Scale. Psychol Assess. 28, 1452–1464. DOI: 10.1037/pas0000276

PT

Gray, M. J., Litz, B. T., Hsu, J. L. Lombardo, T. W., 2004. The psychometric properties of the Life Events Checklist. Assessment. 11, 330–341. DOI: 10.1177/103191104269954

CE

Hooley, J. M., Franklin, J. C. 2018. Why do people hurt themselves? A new conceptual model of nonsuicidal self-injury. Clin Psychol Sci. 6, 428–451.

AC

Hu, L., Bentler, P. M., 1999. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Model. 6, 1–55.

Joiner, T. E., 2005. Why People Die by Suicide. Cambridge, MA: Harvard University Press. Joiner, T. E., Pfaff, J. J., Acres, J. G., 2002. 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. Behav Res Ther. 40, 471–481.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

29

Keshaviah, A., Edkins, K., Hastings, E. R., Krishna, M., Franko, D. L., Herzog, D. B., Thomas, J. J., Murray, H. B., Eddy, K. T., 2014. Re-examining premature mortality in anorexia nervosa: a meta-analysis redux. Compr Psychiatry. 55, 1773–1784. DOI: 10.1016/j.comppsych.2014.07.017 Klonsky, E. D., 2011. Non-suicidal self-injury in United States adults: Prevalence, sociodemographics,

CR IP T

topography and functions. Psychol Med. 41, 1981–1986. DOI: 10.1017/S0033291710002497. Klonsky, E. D., May, A. M., 2015. The three-step theory (3ST): A new theory of suicide rooted in the “ideation-to-action” framework. Int J Cogn Ther. 8, 114–129. DOI: 10.1521/ijct.2015.8.2.114 Li, C. H., 2016. Confirmatory factor analysis with ordinal data: comparing robust maximum likelihood

AN US

and diagonally weighted least squares. Behav Res Methods. 48, 936–349. DOI: 10.3758/s13428015-0619-7

Little, T. D., Slegers, D. W., 2005. Factor analysis: multiple groups. In B. S. Everitt, D. C. Howell (Eds.) Encyclopedia of statistics in behavioral science, volume 2, (pp. 617–623). Hoboken, NJ: John

M

Wiley & Sons.

Lloyd, E. E., Kelley, M. L., Hope, T., April 1997. Self-mutilation in a community sample of adolescents:

ED

Descriptive characteristics and provisional prevalence rates. Poster session presented at the annual meeting of the society for Behavioral Medicine, New Orleans, LA.

PT

Morgan, J. F., Reid, F., Lacey, J. H., 1999. The SCOFF questionnaire: assessment of a new screening tool for eating disorders. BMJ. 172, 164–165.

CE

Muthén, L. K., Muthén, B. O., 1998–2012. Mplus User’s Guide. Seventh Edition. Los Angeles, CA: Muthén & Muthén.

AC

Nock, M. K., Joiner, T. E., Gordon, K. H., Lloyd-Richardson, E., Prinstein, M. J., 2006. Non-suicidal self-injury among adolescents: diagnostic correlates and relation to suicide attempts. Psychiatry Res. 144, 65–72. DOI: 10.1016/j.psychres.2006.05.010

O’Connor, R. C., 2011. Towards an integrated motivational-volitional model of suicidal behaviour. In O’Connor, R. C., Platt, S., Gordon, J. (Eds.), International handbook of suicide prevention: Research, policy, and practice (pp. 181–198). Malden, MA: John Wiley & Sons.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

30

Poindexter, E. K., Nazem, S., Forster, J. E., 2017. Painful and provocative events scale and fearlessness about death among Veterans: Exploratory factor analysis. J Affect Disord. 208, 528–534. DOI: 10.1016/j.jad.2016.10.030 Ribeiro, J. D., Franklin, J. C., Fox, K. R., Bentley, K. H., Kleiman, E. M., Chang, B. P., Nock, M. K.,

CR IP T

2015. Self-injurious thoughts and behaviors as risk factors for future suicide ideation, attempts, and death: a meta-analysis of longitudinal studies. Psychol Med. 46, 225–236. DOI: 10.1017/S0033291715001804

Ribeiro, J. D., Witte, T. K., Van Orden, K. A., Selby, E. A., Gordon, K. H., Bender, T. W., Joiner, T. E.,

AN US

2014. Fearlessness about death: The psychometric properties and construct validity of the revision to the Acquired Capability for Suicide Scale. Psychol Assess. 26, 115-126. DOI: 10.1037/a0034858

Rokke, P. D., Fleming-Ficek, S., Simens, N. M., Hegstand, H. J., 2004. Self-efficacy and choice of

M

coping strategies for tolerating acute pain. J Behav Med. 27, 343-360. Sass, D. A., 2011. Testing measurement invariance and comparing latent factor means within a

ED

confirmatory factor analysis framework. J Psychoed Assess. 29, 347–363. DOI: 10.1177/0734282911406661

PT

Selby, E. A., Anestis, M. D., Joiner, T. E., 2007. Daydreaming about death: Violent daydreaming as a form of emotion dysregulation in suicidality. Behav Modif. 31, 867–879. DOI:

CE

10.1177/0145445507300874 Shapiro, D. N., Chandler, J., Mueller, P. A., 2013. Using Mechanical Turk to study clinical populations.

AC

Clin Psychol Sci. 1, 213–220. DOI: 10.1177/2167702612469015

Smith, A. R., Ribeiro, J. D., Mikolajewski, A., Taylor, J., Joiner, T. E., Iacono, W. G., 2012. An examination of environmental and genetic contributions to the determinants of suicidal behavior among male twins. Psychiatry Res. 197, 60–65. DOI: 10.1016/j.psychres.2012.01.010

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

31

Smith, P. N., Cukrowicz, K. C., 2010. Capable of suicide: A functional model of the acquired capability component of the Interpersonal–Psychological Theory of Suicide. Suicide Life Threat Behav. 40, 266–275. DOI: 10.1521/suli.2010.40.3.266 Smith, P. N., Selwyn, C., D’Amato, D., Granato, S., Kuhlman, S., Mandracchia, J. T., 2016. Life

CR IP T

experiences and the acquired capability for suicide in incarcerated men. Death Stud. 40, 432–439. DOI: 10.1080/07481187.2016.1171264

Tabachnick, G. G., Fidell, L. S., 2001. Using multivariate Statistics. Boston, MA: Allyn and Bacon. Teismann, T., Forkmann, T., Wachtel, S., Edel, M.-A., Nyhuis, P., Glaesmer, H., 2015. The German

AN US

version of the Painful and Provocative Events Scale: A psychometric investigation. Psychiatry Res. 226, 264–272. DOI: 10.1016/j.psychres.2014.12.060

Templin, J., 2017. Testing multi-group measurement invariance using WLSMV in item factor models in Mplus version 7.4 [PDF document]. Retrieved from https://jonathantemplin.com/latent-trait-

M

measurement-structural-equation-modeling-fall-2017-epsy-906-ku/#1503074103352-a065ff7f9da0.

ED

Van Orden, K. A., Cukrowicz, K. C., Witte, T. K., Braithwaite, S. R., Selby, E. A., Joiner, T. E. Jr., 2010. The interpersonal theory of suicide. Psychol Rev. 117, 575–600. DOI; 10.1037/a0018697

PT

Van Orden, K. A., Cukrowicz, K. C., Witte, T. K., Joiner, T. E., 2012. Thwarted belongingness and perceived burdensomeness: Construct validity and psychometric properties of the Interpersonal

CE

Needs Questionnaire. Psychol Assess. 24, 197-215. DOI: 10.1037/a0025358 Van Orden, K. A., Witte, T. K., Gordon, K. H., Bender, T. W., Joiner, T. E., 2008. Suicidal desire and the

AC

capability for suicide: tests of the interpersonal–psychological theory of suicidal behavior among adults. J Consult Clin Psychol. 76, 72–83. DOI: 10.1037/0022-006X.76.1.72

Whiteside, S. P., Lynam, D. R., 2001. The five factor model and impulsivity: Using a structural model of personality to understand impulsivity. Pers Individ Dif. 30, 669–689. DOI: 10.1016/S01918869(00)00064-7

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

32

Witte, T. K., Correia, C. J., Angarano, D., 2013. Experience with euthanasia is associated with fearlessness about death in veterinary students. Suicide Life Threat Behav. 43, 125–138. DOI: 10.1111/sltb.12000 Witte, T. K., Gordon, K. H., Smith, P. N., Van Orden, K. A., 2012. Stoicism and sensation seeking: male

CR IP T

vulnerabilities for the acquired capability for suicide. J Res Pers. 46, 384–392. DOI: 10.1016/j.jrp.2012.03.004

Witte, T. K., Zuromski, K. L., Smith, A. R., Bodell, L. P., Goodwin, N., Siegfried, N., Bartlett, M. L., 2016. Restrictive eating: associated with suicide attempts, but not acquired capability in

AN US

residential patients with eating disorders. Psychiatry Res. 235, 90–96. DOI: 10/1016/j.psychres.2015.11.043

World Health Organization (2015). Suicide Fact Sheet. Retrieved from http://www.who.int/mediacentre/factsheets/fs398/en/.

M

Zuckerman, M., Eysenck, S. B., Eysenck, H. J., 1978. Sensation seeking in England and America: Crosscultural, age, and sex comparisons. J Consult Clin Psychol. 46, 139–149.

ED

Zuromski, K. L., Cero, I., Witte, T. K., 2018). Non-monotonic temporal variation in fearlessness about death: A latent class growth analysis. Psychiatry Res. 268, 46–52. DOI:

AC

CE

PT

10.1016/j.psychres.2018.06.057.

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE Table 1 Demographic characteristics for the Study 1 (exploratory factor analysis) and Study 2 (confirmatory factor analysis) samples

181 (44.9) 222 (55.1) 0 (0)

370 (82.8) 29 (6.5) 25 (5.6) 5 (1.1) 1 (0.2) 17 (3.8)

327 (81.1) 34 (8.4) 23 (5.7) 4 (1.0) 1 (0.2) 14 (3.4)

34 (7.6) 413 (92.4) M (SD) 35.4 (12.0)

36 (8.9) 367 (91.1) M (SD) 38.1 (12.6)

ED PT CE AC

CR IP T

205 (45.9) 239 (53.5) 3 (0.6)

M

Age

Study 2 (N = 403) n (%)

AN US

Gender Men Women Non-binary or other Race Caucasian African American or Black Asian American Indian/Alaska Native Hawaiian or Other Pacific Islander Multiracial Ethnicity Hispanic or Latino Not Hispanic or Latino

Study 1 (N = 447) n (%)

33

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

34

Table 2 Factor loadings from the Geomin-rotated final two-factor solution of the Painful and Provocative Events Scale–Revised, identified through exploratory factor analysis Item

Factor

AC

CE

PT

ED

M

AN US

CR IP T

1 *4. Have you gone on a motorcycle? .49 *5. Have you shot a gun? .48 *6. Have you broken a bone? .58 *7. Have you been in a car accident? .48 *8. Have you had injuries requiring medical attention? .63 11. Have you seen dead or mutilated body parts? .43 13. Have you had consensual unprotected sex with someone you did not know very well? .44 15. Have you ever been in a fire (in a burning building)? .54 16. Did your parents use corporal punishment? .35 17. Have you ever torn a ligament or tendon? .55 18. Have you ever seen someone die? .67 20. Has someone close to you died by suicide? .43 21. Have you ever had a root canal? .44 22. Have you seen someone else receive a serious injury that required medical attention? .53 23. Have you seriously injured yourself by accident (e.g., burning your hand while cooking, which required medical attention)? .52 25. Have you had an extensive, painful surgery? .44 26. Have you accidentally cut a finger while cooking, severely enough to require stitches? .56 1. Have you made yourself vomit? .06 2. Have you taken diet pills? .16 3. Have you gone eight or more hours while you were awake without eating? .27 * 9. I have stopped myself right before attempting suicide and did not carry out the attempt (e.g., standing on a bridge but not jumping, loading a gun but not pulling the trigger). .09 10. Have you had impulses to hurt yourself (without actually hurting yourself)? –.13 12. Have you spent time thinking about your own death in detail? .09 14. Have you had violent daydreams or fantasies? .00 19. Have you ever engaged in deliberate self-harm (e.g., cutting or burning your skin)? –.05 24. Have you collected items in preparation to seriously harm yourself (e.g., collected pills, razors, etc.)? .01 Note. *Indicates an item from the original PPES. Bold typeface indicates items that load primarily onto the given factor.

2 –.14 –.02 .00 .08 .17 .15 .20 –.23 .16 –.07 –.18 .03 –.13 .10 .26 –.06 .07 .49 .37 .45 .73 .96 .70 .57 .91 .87

REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE Table 3

CR IP T

ACCEPTED MANUSCRIPT

35

Correlations between factors and indicators of convergent and discriminant validity in the Study 1 (exploratory factor analysis) sample (N = 447) 1 — .32** .31** .21** .19** .06 .02 .15** 29.88 7.65 0.63 0.48

2

3

4

5

— .24** .12** .04 .38** .34** .50** 17.04 6.70 0.99 0.28

— .25** .28** –.01 –.08 .10* 51.22 21.06 0.01 0.36

— .28** .01 –.10 .14** 14.53 7.42 –0.89 –0.83

— .00 .05 .06 33.64 12.27 0.13 –0.78

CE

PT

ED

M

Note. PPES-R = Painful and Provocative Events Scale–Revised, * p < .05, ** p < .001.

AC

6

AN US

1. PPES-R: Actual or potential injuries 2. PPES-R: Self-harm 3. Pain tolerance 4. Fearlessness about death 5. Sensation seeking 6. Perceived burdensomeness 7. Thwarted belongingness 8. Suicidal ideation Mean SD Skew Kurtosis

— .65** .59** 3.72 7.11 2.24 4.47

7

8

— .46** 22.69 10.25 0.77 –0.37

— 0.61 1.52 2.70 7.11

REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE Table 4

CR IP T

ACCEPTED MANUSCRIPT

Fit statistics and tests of measurement invariance in the Study 2 (confirmatory factor analysis) sample (N = 403)

AC

CE

PT

ED

M

AN US

χ2 df χ2 difference RMSEA (90% CI) CFI TLI 2-factor model 861.00** 298 n/a .068 (.063, .074) .905 .897 Configural invariance 1181.25** 596 n/a .070 (.064, .076) .925 .918 Metric invariance 1230.55** 620 84.38** .070 (.064, .076) .922 .918 Scalar invariance 1369.68** 715 176.68** .067 (.062, .072) .917 .925 Note. χ2 difference indicates the result from the DIFFTEST option in Mplus, n/a = not applicable, RMSEA = root mean-square error of approximation, CI = confidence interval, CFI = comparative fit index, TLI = Tucker–Lewis index, * p < .05, ** p < .001.

36

REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE Table 5

CR IP T

ACCEPTED MANUSCRIPT

37

Correlations among factors and indicators of convergent and discriminant validity in the Study 2 (confirmatory factor analysis) sample (N = 403) 6

7

AC

CE

PT

ED

M

AN US

1 2 3 4 5 1. PPES-R: Actual or potential injuries — 2. PPES-R: Self-harm .60** — 3. Pain tolerance .22** .18** — 4. Fearlessness about death .19** .06 .32** — 5. Sensation seeking .24** .15** .35** .20** — 6. Nonsuicidal self-injury methods .30** .50** .13** .01 .10* 7. Disordered eating .21** .45** .02 –.04 .07 8. Traumatic life events .64** .50** .22** .12* .11* 9. Perceived burdensomeness .20** .44** .09 –.05 .23** 10. Thwarted belongingness –.05 .23** –.03 –.05 .00 11. Suicidal ideation .32** .60** .12* .05 .15** Mean 13.55 6.55 2.06 14.38 32.02 SD 10.40 6.44 1.21 7.36 12.73 Skew 1.08 1.49 –0.16 0.01 0.09 Kurtosis 1.17 2.34 –0.87 –0.68 –1.08 Note. PPES-R = Painful and Provocative Events Scale–Revised, * p < .05, ** p < .001.

— .33** .31** .31** .18** .47** 0.88 1.51 2.01 3.73

— .23** .33** .23** .32** 0.73 1.13 1.57 1.77

8

9

10

11

— .19** .05 .26** 3.36 2.99 1.37 2.28

— .49** .54** 4.56 8.14 1.83 2.23

— .29** 15.21 12.94 0.58 –0.62

— 0.60 1.73 3.45 12.55

ACCEPTED MANUSCRIPT REVISING THE PAINFUL AND PROVOCATIVE EVENTS SCALE

38

Figure 1 Standardized confirmatory factor analysis results and correlations between the two Painful and Provocative Events Scale–Revised latent factors (Study 2, N = 403)

PPES-R 4. Gone on a motorcycle

.55 .69

PPES-R 5. Shot a gun PPES-R 6. Broken a bone

CR IP T

.58

PPES-R 7. Been in a car accident

.57 .74 .75

PPES-R 11. Seen dead or mutilated body parts

.55

PPES-R 13. Had consensual unprotected sex with stranger

.74

PPES-R 15. Been in a fire (in a burning building)

.50

PPES-R 16. Parents used corporal punishment

AN US

Actual or potential injuries

PPES-R 8. Injuries requiring medical attention

.69

PPES-R 17. Torn a ligament or tendon

.68

.72

PPES-R 18. Seen someone die

.45

PPES-R 20. Someone close died by suicide

.73

PPES-R 21. Had root canal

.70

M

.66

PT

ED

.71

AC

CE

Self-harm

.67

PPES-R 22. Seen someone receive serious injury PPES-R 23. Seriously injured self by accident

PPES-R 25. Had extensive, painful surgery PPES-R 26. Accidentally cut a finger while cooking

.58

PPES-R 1. Made yourself vomit

.58

PPES-R 2. Taken diet pills

.56

PPES-R 3. Gone 8+ hours while awake without eating

.94

PPES-R 9. Stopped self right before attempting suicide

.85

PPES-R 10. Had impulses to hurt self

.70

PPES-R 12. Spent time thinking about own death .70 .84 .92

PPES-R 14. Had violent daydreams or fantasies PPES-R 19. Engaged in deliberate self-harm PPES-R 24. Collected items in preparation to harm self

Note. PPES-R = Painful and Provocative Events Scale–Revised.