An ecological momentary assessment investigation of complex and conflicting emotions in youth with borderline personality disorder

Psychiatry Research 252 (2017) 102–110

Contents lists available at ScienceDirect

Psychiatry Research journal homepage: www.elsevier.com/locate/psychres

An ecological momentary assessment investigation of complex and conflicting emotions in youth with borderline personality disorder

MARK

Holly E. Andrewesa,b, Carol Hulberta, Susan M. Cottonb,c, Jennifer Bettsb,c, Andrew ⁎ M. Chanenb,c,d, a

Melbourne School of Psychological Sciences, The University of Melbourne, Melbourne, Australia Orygen, The National Centre of Excellence in Youth Mental Health, Melbourne, Australia c Centre for Youth Mental Health, The University of Melbourne, Melbourne, Australia d Orygen Youth Health, NorthWestern Mental Health, Melbourne, Australia b

A R T I C L E I N F O

A BS T RAC T

Keywords: Borderline personality disorder Emotional acceptance Non-suicidal self-injury Youth Psychiatry

Non-suicidal self-injury (NSSI) is a prevalent behaviour among people with borderline personality disorder (BPD) but many aspects of the emotional changes that trigger and maintain this behaviour are unknown. This study examines the relationships between NSSI and the number of negative (‘negative complex’) and opposing valence (‘conflicting’) emotions. One hundred and seven youth (aged 15–25 years) with first-presentation BPD were assessed using a combination of self-report and ecological momentary assessment to investigate trait levels of emotional acceptance and in vivo changes in the number of negative complex and conflicting emotions before and after self-injurious thoughts and behaviours. Multilevel modelling revealed that changes in the number of negative complex emotions mirrored distress levels before and after self-injurious thoughts and behaviours, approximating a quadratic curve. Increases in the number of negative complex emotions reported prior to selfinjurious thoughts and behaviours were associated with lower acceptance of negative emotions. These findings indicate that the number of negative emotions experienced contributes to distress prior to engagement in NSSI. The relationship between non-acceptance of negative emotions and negative complex emotions prior to NSSI suggests that improved emotional awareness and acceptance should be a focus for early interventions aimed at reducing self-injury.

1. Introduction Emotion dysregulation, broadly defined as the inability to flexible respond to and manage emotions (Carpenter and Trull, 2013), is a core feature of Borderline Personality Disorder (Conklin et al., 2006; EbnerPriemer et al., 2015; Linehan et al., 2007). Non-suicidal self-injury (NSSI; Chapman et al., 2005; Klonsky, 2007) is a behaviour commonly used to regulate emotions in BPD, defined as the deliberate destruction of one's bodily tissue that is without cultural significance or lethal intent (Nock, 2009). The relationship between emotion dysregulation and NSSI has been studied in BPD with respect to the intensity of negative emotions and emotional instability. For example, both negative affect (Andrewes et al., 2016; Chapman and Dixon-Gordon, 2007; Kleindienst et al., 2008; Reitz et al., 2012), perceived rejection and dissociation (Snir et al., 2015) have been found to increase prior to NSSI and to reduce following NSSI. Emotional instability has also been identified as a predictor of NSSI in undergraduate (Selby et al., 2013) and eating disorder populations (Anestis et al., 2012; Vansteelandt ⁎

et al., 2013). Less is known, however, about the quantity of emotions experienced prior to NSSI and how other maladaptive strategies, such as non-acceptance or suppression of emotions, might contribute to the experience of multiple emotions during this time. This knowledge gap limits our understanding of the emotional experiences that contribute to engagement in NSSI. Accordingly, the focus for the current study is the relationship between multiple emotional experiences, non-acceptance of emotions and NSSI in youth with BPD. 1.1. Relationship between multiple emotional experiences, distress and NSSI Multiple emotional experiences, also termed ‘complex’ emotions, have been associated with increased distress in three studies investigating non-BPD populations (Choi et al., 2015; Eatough et al., 2008; Pearson et al., 2008). In the first, women from a community sample described experiencing multiple emotions during expressions of aggressive behaviour (Eatough et al., 2008). In two other studies of

Correspondence to: Orygen, the National Centre of Excellence in Youth Mental Health, Locked Bag 10, Parkville, Victoria 3052, Australia. E-mail address: [email protected] (A.M. Chanen).

http://dx.doi.org/10.1016/j.psychres.2017.01.100 Received 29 August 2016; Received in revised form 18 December 2016; Accepted 12 January 2017 Available online 27 February 2017 0165-1781/ © 2017 Elsevier B.V. All rights reserved.

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

chronically depressed adult outpatients, multiple emotional experiences were associated with increased rumination (Pearson et al., 2008) and a reduced ability to resolve self-criticism (Choi et al., 2015). Only two publications have investigated multiple emotions and their association with distress in BPD (Ebner-Priemer et al., 2008, 2007). These publications relate to one experiment, in which adults with BPD (n = 50; Mage = 31.3 years) and healthy controls (n = 50; Mage = 27.7 years) identified their primary and secondary emotions from a provided list every 10- to 20-min over a 24-h period. These data were acquired using experience sampling or ecological momentary assessment (EMA), which has become the preferred method of identifying affect changes surrounding NSSI (Armey, 2012) as it largely obviates retrospective bias and maintains ecological validity (Shiffman et al., 2008). In these studies, adults with BPD exhibited multiple negative emotions (termed ‘negative complex emotions’), both more commonly and at a higher intensity than healthy controls (Ebner-Priemer et al., 2007). Un-medicated adults with BPD (n = 10) also exhibited a higher heart rate and increased vagal activity compared with healthy controls, indicating both sympathetic and parasympathetic nervous system activation (Ebner-Priemer et al., 2007). Despite the ambiguous relationship between negative complex emotions and distress found in the latter study, reports from a case series describe people with BPD as experiencing an ‘avalanche’ of emotions, such as internalised and externalised anger, guilt and sadness, prior to engaging in NSSI (Leibenluft et al., 1987). This suggests that people with BPD experience negative complex emotions prior to engagement in NSSI. Ebner-Priemer and colleagues (2008) also investigated multiple emotions of opposing valences, named ‘conflicting emotions’, finding that their BPD group reported this emotional experience more frequently than did healthy controls, and that their presence was related to self-reported psychological distress in both groups (measured via a 10-point Likert scale of distress). The finding that multiple emotional experiences occur during distress provides partial support for Linehan's (1993, 2007) theory of ‘secondary emotions’ for BPD. In this theory, multiple emotional experiences (‘secondary emotions’) are posited to occur in response to a failure to accept, and a desire to suppress, a primary emotion. The experience of secondary emotions is theorised to amplify and prolong experiences of distress and to increase the odds of engaging in a maladaptive behaviour, such as NSSI. Yet, to date, the relationship between multiple emotional experiences (‘negative complex’ and ‘conflicting’ emotions) and distress has not been substantiated by the assessment of distress-related behaviours, such as the occurrence of self-injurious thoughts (SIT) or NSSI. Also, the generalisability of the aforementioned research (Ebner-Priemer et al., 2008, 2007) is limited by the number of participants who were undergoing Dialectical Behaviour Therapy (DBT; 42%) at the time of the study. The emotion regulation skills training component of DBT potentially modified these participants’ naturally occurring experiences of ‘negative complex’ and ‘conflicting’ emotions, along with the associated levels of distress. The current study aims to address the limitations of prior research by investigating the relationship between multiple emotions, distress and NSSI early in the course of the disorder and prior to the effects of BPD specific psychotherapy, long-term polypharmacy, or entrenched functional problems that might modulate emotion regulation skills (Chanen, 2015). As the expression and understanding of negative complex and conflicting emotions develops prior to 10 years of age in normally developing children, multiple emotional experiences should be present early in the developmental course of BPD (Harter, 1983; Witre and Vallance, 1994).

as posited in Linehan's (1993, 2007) theory of 'secondary emotions', is yet to be empirically investigated. Further, limited investigations of the relationship between non-acceptance of emotions and distress or NSSI in BPD have been conducted and the findings from these studies are mixed. For example, female outpatients with BPD who engaged in a 14week group aimed at improving emotional acceptance (Gratz and Gunderson, 2006) experienced enhanced emotion regulation skills and reduced acts of self-harm, compared with treatment as usual (Gratz and Gunderson, 2006). Similarly, experiential avoidance, a related construct posited to enhance non-acceptance of emotions (Chapman et al., 2006), has been identified as the most common motive for engaging in NSSI in youth and adults with BPD (Andrewes et al., 2016; Brown et al., 2002; Hulbert and Thomas, 2010). In contrast, adults with BPD from an outpatient and community setting asked to accept their emotional experiences acknowledged a greater urge to engage in NSSI and self-punishment than participants asked to suppress them (n =36; Svaldi et al., 2012). Similarly, using an experiential sampling design, undergraduate students with elevated levels of BPD pathology who were asked to accept or observe their negative emotional experiences reported a greater number of urges to engage in impulsive behaviour and lower number of positive emotions, compared with days they were asked to supress them. There were no significant differences, however, in the number of negative emotions experienced between days when asked to accept and suppress emotions (Chapman et al., 2009). These conflicting findings indicate that further research is required to clarify the relationship between acceptance of emotions and distress. Furthermore, an investigation of the relationship between non-acceptance of difficult emotions and multiple emotional experiences during distress, and in the context of NSSI, remains a gap in the literature. Employing a youth sample of BPD with limited exposure to treatment also provides insight into these relationships early in the course of the disorder and prior to BPD specific psychological treatment which might alter emotion regulation skills.

1.2. Non-acceptance of emotions and its relationship with multiple emotions, distress and NSSI

One-hundred and thirteen acutely unwell, treatment-seeking patients with first-presentation BPD were recruited from two government-funded mental health services in western metropolitan Melbourne, Australia, as part of a larger randomised control trial (see Chanen et al., 2015). Six participants were excluded due to failure

1.3. The current study: aims and hypotheses Limited and often conflicting findings from studies investigating the relationship between multiple emotions, non-acceptance of difficult emotions and distress constrains our understanding of the emotional experiences that contribute to engaging in NSSI. The current study aims to investigate the role of ‘negative complex’ and ‘conflicting’ emotions and non-acceptance of emotions in the context of engaging in self-harm and self-injurious thoughts, which substantiate the experience of distress in youth with BPD. In the absence of prior or conflicting empirical research, hypotheses were informed by Linehan's theory of 'secondary emotions' for BPD (1993; 2007). It was predicted that (i) an increase in the number of negative complex and conflicting emotions would occur prior to SIT and NSSI, with a reduction following SIT and NSSI that mirrors participants’ changing levels of distress and fits a quadratic curve; (ii) for participants who did not engage in SIT or NSSI, changes in the number of negative complex and conflicting emotions and distress levels would not fit a quadratic curve; and (iii) lower acceptance of negative emotions would be associated with an increase in ‘negative complex’ and ‘conflicting emotions’ prior to SIT and NSSI. 2. Methods 2.1. Participants

The relationship between a failure to accept emotions, the experience of multiple secondary emotions, distress and engagement in NSSI 103

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

10 pm. Participants were given 15-min from the signal to commence the survey and a total of 8-min to complete the survey before they were locked out. If participants failed to complete at least one survey during a day, the phone was left with them for another six days. Participants attending school were provided with information for their teachers to contact the study organisers, should they wish to respond to prompts during class time. An automatic alert system was activated when participants responded in a manner that indicated high-risk according to a priori thresholds. On these occasions a clinician received an alert and called the participant to complete a preliminary risk assessment. Participants were reimbursed $40 for their participation.

of the EMA safety alert system or failure to complete any questionnaires. The final sample comprised 107 youth, aged 15–25 years (inclusive), with no prior experience of evidence-based BPD treatment. 2.2. Measures 2.2.1. Non-EMA measures 2.2.1.1. Mental state disorder and personality disorder diagnoses. The Structured Clinical Interview for DSM-IV-TR (SCID) Axis I Disorders – Patient Edition (SCID-I/P; First et al., 2002) and the SCID for Axis II Disorders (SCID-II; (First et al., 1997) were employed to measure mental state and personality disorders respectively.

2.4. Statistical analysis 2.4.1. Preliminary analysis An assessment of normality, linearity, homoscedasticity, or univariate and multivariate outliers was performed for all variables in the study (Tabachnick and Fidell, 2013). One-way analysis of variance (ANOVA) models and chi-square (χ2) tests for independence were employed to compare the demographics. A χ2 test was used to compare global missing EMA data and missing data from the 2 h pre- and postNSSI or SIT.

2.2.1.2. Trait levels of emotional acceptance. The Difficulties in Emotion Regulation Scale (DERS; Gratz and Roemer, 2004) assesses trait levels of emotion regulation via 36 self-report items. The subscale, ‘non-acceptance of negative emotional responses’ determined the association between non-acceptance and the number of negative complex or conflicting emotions. Two subscales, ‘lack of emotional awareness’ and ‘lack of emotional clarity’, were used to clarify whether a report of multiple emotions indicated lower levels of emotional differentiation or the true experience of negative complex or conflicting emotions. All three scales have high internal consistency (Cronbach's α range 0.80–0.85) and good 4–8 week re-test reliability (0.68–0.80; Gratz and Roemer, 2004).

2.4.2. Statistical analyses for hypotheses 1 and 2 To compare changes in the number of negative complex and conflicting emotions and intensity of distress before and after SIT and NSSI, five consecutive time points were selected by centring time on the NSSI or SIT event, and selecting two time points immediately before and after, and within 20 h of the event. Time was centred on the recording of SIT and NSSI (T0). This timeframe was selected because previous modelling in this sample revealed that changes in negative affect begin a median of 15-h prior to NSSI (Andrewes et al., 2016). Multiple NSSI events occurring over a 24-h period were excluded, as per Snir et al. (2015). As a consequence of the small size of the sample who engaged in NSSI, multiple events could not be analysed separately from single events. For participants who engaged in neither SIT nor NSSI, the same criteria were used for selecting the data points before and after a random point. A random point was chosen rather than a matched time point, because engaging in the SIT and NSSI events occurred relatively randomly throughout the day and over each week. For example, SIT and NSSI occurred on the weekends 22% and 29% of the time, respectively (29% indicates an even distribution). Furthermore, 33.3% of all NSSI events occurred within each 4-h time block throughout the day (10–2 pm, 2–6 pm, 6–10 pm) indicating an even distribution, while 38%, 42% and 23% of SIT occurred between 10 am-2 pm, 2–6 pm and 6–10 pm, respectively. A two-level multilevel model was employed to analyse the EMA data because this method allows the assessment of affect changes within an individual over time (level 1) and between individuals over time (level 2). Maximum likelihood estimation with robust standard errors were employed to adjust for non-normality (Maas and Hox, 2004) and missing data (Yuan and Bentler, 2000). Bootstrapped confidence intervals were used to reduce the influence of sample size. As timing of responses differed for each participant, a variable representing time (in hours) was created, and the intercepts, linear and quadratic slopes were allowed to vary using random effects. Fit statistics were compared before adding a time-invariant covariate to compare the three groups: SIT, NSSI and a group who engaged in neither SIT nor NSSI identified as the comparison group. These were dummy coded, with the comparison group coded 00 as the reference variable (Hox, 2010). To compare the group who engaged in SIT with the group who engaged in NSSI, SIT was coded as 00. The standard multilevel equation for this model, with the comparison group as the reference is: Yij =γ00 +γ10 TIMEij +γ20 TIMEij2 +γ01NSSI +γ02SIT +γ11 (NSSIj×TIMEij) +γ21 (NSSIj×TIMEij2) +γ01SIT+γ11 (SITj×TIMEij) +γ21 (SITj×TIMEij2) ζ0j +ζ1jTIMEij +ζ2jTIMEij2 +εij.

2.2.2. EMA measures 2.2.2.1. Negative complex and conflicting emotions. Multiple emotions were identified via ratings on the 10-item short-form of the Positive and Negative Affect Scale (PANAS; Kercher, 1992), which was delivered through the Mobiletype© program (Reid et al., 2009). This measure consists of 5 positive and 5 negative emotions, which are rated from 1 to 5 (“very slightly or not at all” to “extremely”). The reliability of the subscales is high for community samples aged between 18 and 75 years, with a Cronbach's α between 0.78 and 0.87 (Mackinnon et al., 1999). The structural characteristics of the PANAS are stable across gender and age groups (Mackinnon et al., 1999). Intensity of psychological distress was calculated as the mean rating of the “distress” item. Negative complex emotions were defined as two or more co-occurring negative emotions. Conflicting emotions were defined as the co-occurrence of at least two emotions of opposing valence. Only emotions rated 2 or above were included because this rating reliably indicates that these emotions are present. Emotional experiences reported were scored according to the number of negative or opposing valence emotions occurring simultaneously, with possible ranges of 0–5 and 0–10 respectively. SIT and NSSI events were also captured via a set of closed questions delivered by the Mobiletype© program including: “Since the last signal have you thought about deliberately hurting yourself?”; “Since the last signal, have you actually hurt yourself?”

2.3. Procedure Ethical approval was obtained from the Melbourne Health Human Research Ethics Committee. Written informed consent was obtained from participants (and from a legal guardian for minors). The SCID-I/ P, SCID-II and DERS were completed and then participants were issued with a mobile phone loaded with the Mobiletype© program, that prompted participants six times per day for six days to answer questions regarding their affect and engagement in NSSI and SIT. Prompts were randomised within 2-h time-blocks between 10 am and 104

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

2.4.3. Post-hoc analysis - duration of time between NSSI or SIT engagement and initiation/termination of distress and negative complex and conflicting emotions The duration of time between the initial and final changes in distress and number of negative complex and conflicting emotions occurring prior to and following SIT or NSSI, respectively, was assessed post hoc, over the 6 days of data collection. The time of initial affect change reported prior to NSSI or SIT was identified as the lowest or highest rating of affect, respectively, prior to an increase or reduction leading to NSSI or SIT. The time that affect change terminated following NSSI or SIT was identified as the lowest or highest rating of affect, respectively, prior to an increase or reduction in affect. Participants were excluded from selection if: (i) more than 24-h of missing data were identified in the time point prior to or following NSSI or SIT and (ii) if affect remained unchanged in the 100-h prior to or following NSSI or SIT. For individuals who engaged in more than one NSSI or SIT event, only affect changes occurring prior to the first event were included.

Table 1 Sample characteristics for participants who engaged in self injurious thoughts only (SIT), Non-suicidal self-injury (NSSI) or neither behaviour over the 6 days of data collection (comparison).

Female Age Caucasian Indigenous Australian Low SESd Completed Schoole Unemployed Leave from school/work Employed parttime Employed fulltime Mental state disorder Mood disorders Anxiety Disorders Eating Disorders Personality disorders Antisocialf Paranoid Avoidant Narcissistic Histrionic Dependent

2.4.4. Statistical analyses for hypothesis 3 Pearson Product Moment correlation coefficients (r) were used to identify associations between self-reported emotion regulation skills (e.g., non-acceptance of negative emotions) and the changes in the number of negative complex and conflicting emotions from the response point prior to T0 and at T0. Participants were excluded from analysis if more than 24-h of missing data were identified in the time point prior to T0. Correlations were compared using a Fisher r to zscore transformation. To account for the different timings of data entry prior to T0, the change in the number of negative complex and conflicting emotions was divided by time, with the gradient (slope) of change in complex and conflicting emotions correlated with trait emotion regulation ratings for each participant.

Total (107)

NSSIa (24)

SITb (46)

Comparisonc (37)

83.2% 18.1 (SD=2.7) 91.0% 2.8%

87.5% 18.2 (SD=2.9) 95.8% 4.2%

84.4% 19.0 (SD=2.6) 91.1% 2.2%

78.9% 19.3 (SD=2.8)

45.8% 38.8%

29.2% 75.0%

52.2% 47.4%

48.65% 18.0%

12.0% 49.0%

20.8% 45.8%

10.9% 50.0%

8.1% 48.6%

11.0%

8.3%

6.5%

18.9%

7.0%

8.3%

6.5%

8.1%

83.2% 71.3% 8.4%

83.3% 75.0% 12.5%

82.2% 75.6% 6.7%

94.7% 63.2% 7.9%

31.8% 20.6% 23.4% 3.7% 3.7% 3.0%

37.5% 29.2% 16.7% 0% 4.2% 4.2%

24.4% 17.8% 28.9% 4.4% 4.4% 4.4%

36.8% 18.4% 21.1% 5.3% 2.6% 0%

86.8% 2.6%

Note. a NSSI, Group engaging in non-suicidal self-injury. b SIT, Group engaging in self-injurious thoughts only. c Comparison, Group engaging in neither NSSI nor SIT. d Socio-economic Status, rated according to the participants’ residential postcode (Vinson, 2007). e Participants over 18 years who completed their final year of schooling. f Diagnosis made ignoring criterion B that requires > 18 years of age; Chanen et al. (2007)).

2.4.5. Measures of effect and statistical programmes employed Measures of effect were calculated for all statistics. Partial eta squared (η2) and Pearson coefficients (r) were calculated for group comparisons and associations, respectively. Local effect sizes for multilevel modelling were identified by the proportional reduction in the variance of the quadratic model after SIT and NSSI were added as predictors (Peugh, 2010). Multilevel modelling was completed using MPLUS version 7.2. All other analysis was completed using IBM® SPSS® Statistics Version 21.

intention and severity (Guertin et al., 2001; Muehlenkamp and Gutierrez, 2007). In the final data set, 24 participants engaged in 52 acts of NSSI (M =2.16, SD =1.51) and 201 SIT (M =8.38, SD =6.74). Forty-six participants engaged in SIT only, with a total of 129 thoughts (M =20.80, SD =2.60), leaving 37 participants in a comparison group, who engaged in neither SIT nor NSSI.

3. Results 3.1. Characteristics of the sample and EMA data

3.2. Negative complex and conflicting emotions before and after NSSI and SIT (Hypothesis 1 and 2)

Non-significant differences between participants who engaged in SIT, NSSI and the comparison group were found for age, gender, social disadvantage status, ethnicity, work status and co-morbidity (see Table 1). Participants completed a total of 1986 diary entries, which equated to an average of 18.56 (SD =10.00) of the possible 36 diary entries were completed (51.56% adherence). Twenty-two participants (20.56%) were given the phone for another 6 days because they completed an insufficient number of data entries in the first 6 days. Independent ttests revealed that participants who received the phone for two weeks did not differ from those who received it for a week on the basis of the demographic variables measured in Table 1 or on the mean number of NSSI or SIT events recorded per participant. A chi-square comparison revealed a non-significant relationship between group membership and the total number of missing entries (χ2 (2, N=107) =8.00, p=0.433) or the number of missing entries in the two hours prior to and following T0 (χ2 (2, N=107) =4.15, p=0.126). Two acts of self-injury with suicidal intent were removed from the sample and excluded from further analysis as these acts differ from NSSI acts in frequency,

The intercept model revealed that 41% of the variance in ratings of distress, 51% of the variance in the number of negative complex emotions and 30% of the variance in the number of conflicting emotions were due to between participant differences that might be predicted by NSSI or SIT. Examination of log-likelihood deviance statistics (Singer and Willett, 2003) showed that change in the number of negative complex emotions, and distress levels before and after T0 fitted a quadratic model significantly better than a linear model (negative complex emotions, χ2Δ (5) =26.22, p < 0.001, distress,χ2Δ (5) =18.27, p=0.003). A cubic model showed no better fit than the quadratic model (negative complex emotions, χ2Δ (2) =0.09, p=0.956 distress, χ2Δ (2) =1.00, p=0.607). The log-likelihood deviance statistic for conflicting emotions revealed that a linear model fitted significantly better than the quadratic model (χ2Δ (5) =−7.50, p < 0.001). Visual inspection of the mean plots (see Fig. 1a and b) supported these findings. Table 2 depicts the multilevel modelling results for a conditional 105

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

5.0 NSSI SIT Comparison

Intensity of Distress

4.0

3.0

2.0

1.0

Mean Number of Negative Complex Emotions

5.0

0.0 -20

-10

0

10

NSSI SIT Comparison

4.0

3.0

2.0

1.0

0.0 -20

20

Timing of Response (hours)

-10 0 10 Timing of Response (hours)

20

a. Mean of ratings of intensity in distress (left) and number of negative complex emotions

Mean Number of Conflicting Emotions

(right) with standard error bars, before and after SIT, NSSI or a random point (comparison group).

10.0

NSSI SIT Comparison

9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 -20

-10 0 10 Timing of Response (hours)

20

b. Mean number of conflicting emotions with standard error bars, before and after SIT, NSSI or a random point (comparison group). Fig. 1. a. Mean of ratings of intensity in distress (left) and number of negative complex emotions (right) with standard error bars, before and after SIT, NSSI or a random point (comparison group). b. Mean number of conflicting emotions with standard error bars, before and after SIT, NSSI or a random point (comparison group).

distress after adding NSSI and SIT as predictors. Only significant random effects were reported. In response to these findings, a post-hoc model which included affect change occurring in the 10-h pre- and post-SIT for distress (Estimate (SE) =−6.27(4.56), z=−1.38, p=0.169, 95% CI [−15.20, 2.66]) and negative complex emotions (Estimate (SE) =−2.29(5.24), z=−0.43, p=0.661, 95% CI [−12.55, 7.96]) was conducted, yet also failed to fit a quadratic curve. Local effect sizes for this model revealed a reduction in the intercept and quadratic slope variance by 19.43% and 83.49% respectively for negative complex emotions and by 20.54% and 43.49% for intensity of distress, after adding SIT as a predictor. An assessment of the linear model for conflicting emotions showed that the average participant who engaged in NSSI (Estimate (SE) =0.91(0.66), z=1.40, p=0.163, 95% CI [−0.37, 2.19]) and SIT (Estimate (SE) =0.70(0.64), z=1.09, p=0.276, 95% CI [−0.56, 1.96]) did not identify higher levels of conflicting emotions than the comparison group at the corresponding time point. Similarly, there were no differences between the number of conflicting emotions reported at

quadratic model estimating changes in distress and negative complex emotions before and after NSSI, SIT and a random point (comparison group). At the point when NSSI and SIT (T=0) were reported, the average participant who engaged in SIT and NSSI identified significantly higher distress levels and a higher number of negative complex emotions than was reported by the comparison group at the corresponding time point. Non-significant differences were found between the number of co-occurring negative emotions, and ratings of distress at T0 for the group who engaged in NSSI and SIT. For participants who engaged in NSSI, the change in the number of negative complex emotions and distress ratings significantly approximated a quadratic curve. The mean curve of negative complex emotions and distress ratings prior to SIT did not fit a quadratic curve. (see Fig. 1a and b). Similarly a quadratic change in negative complex emotions and distress ratings were not found for participants in the comparison group. Local effect sizes revealed that the variance in the intercept and quadratic slope reduced by 24.09% and 73.81% respectively for negative complex emotions and by 21.23% and 38.80% respectively for intensity of 106

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

Table 2 Parameter estimates for the quadratic growth model representing distress levels and the number of complex negative emotions as a function of engagement in NSSI and SIT. Distress Intensity Fixed effectsb Comparison group as reference Intercept NSSI SIT TIME2 NSSIj x TIMEij2 SITj x TIMEij2 SIT group as reference Intercept Δ Intercept NSSI (γ01) SIT x TIME2 (γ20) NSSI x TIMEij2 (γ21) Random effects Level 2 Intercept variance Level 1 Residual variance (ε)

Negative Complex Emotions

Parameter

Estimate (SE)

z

[95% (CI)]a

Estimate (SE)

z

[95% (CI)]a

γ00 γ01 γ02 γ20 γ21 γ22

1.74 (0.28) 1.10 (0.38) 0.72 (0.21) 0.18 (0.67) −1.57 (0.76) −0.60 (0.69)

6.12*** 2.87*** 2.43** 0.27 −2.07** −0.86

[1.18, 2.20] [0.35, 1.73] [0.14, 1.20] [−1.13, 1.29] [−3.05, −0.33] [−1.96, 0.55]

1.29 (0.37) 1.86 (0.63) 1.37 (0.48) 0.45 (0.67) −2.85 (1.00) −0.99 (0.81)

3.41*** 2.93** 2.90** 0.67 −2.85*** −1.23

[0.56, 2.01] [0.72, 3.00] [0.44, 2.30] [−0.86, 1.77] [−4. 809, −0.90] [−2.57, 0.59]

γ01 γ20 γ21 Parameter

2.45 (0.19) 0.38 (0.32) −0.42 (0.42) −0.97 (0.58) Estimate (SE)

12.66*** 1.19 −0.97 −1.65 z

[2.07, 2.83] [−0.25, 1.01] [−1.25, 0.42] [−0.76, 1.74] [95%(CI)] a

2.65 (0.24) 0.49 (0.44) 0.99 (0.81) −1.86 (0.87) Estimate (SE)

11.05*** 1.11 1.23 −2.14* z

[2.18, 2.05] [−0.37, 1.35] [−1.31, 0.23] [−3.56, −0.16] [95%(CI)]

ζ0i

0.77 (0.23)

3.34**

[0.32, 1.15]

1.86 (0.63)

2.93*

[0.62, 3. 10]

εi

1.01 (0.09)

10.87**

[0.83, 1.17]

1.65 (0.13)

12.71**

[1.40, 1.91]

Note. a = Bootstrapped confidence intervals; b = Linear terms were excluded due to non-significance; cNSSI, non-suicidal self-injury; 4SIT, Self-injurious thoughts; *p≤0.050; **p≤0.005; ***p≤0.001; As a result of the discrepancy between group sizes, the variance within the comparison group and SIT group might have obscured the within-person variance in the NSSI+ group. For this reason, caution is required when interpreting the random effects.

from the time point prior to SIT and NSSI. The difference in this correlation for the group who engaged in NSSI, compared with the comparison group, was significant (z=2.15, p=0.032). However, for the SIT group this correlation was not significantly different to that found for the comparison group (z=1.14, p=0.254) or the NSSI group (z=1.25, p=0.211). Correlations between self-reported lack of emotional awareness, clarity, and the increase in negative complex emotions prior to engagement in NSSI and SIT were non-significant. In the comparison group, a significant negative correlation was found between scores on the lack of emotional awareness scale and the number of negative complex emotions. This was not significantly greater than the correlations found for NSSI (z=−0.89, p=0.373) and SIT (z=−1.66, p=0.097). Correlations between all emotion indices and changes in conflicting emotions prior to engagement in SIT, NSSI or a random point were non-significant (see Table 3).

T=0 when the NSSI group (Estimate (SE) =0.21(0.61), z=0.35, p=0.728, 95% CI [−0.99, 1.41]) was compared with the SIT group. The change in conflicting emotions did not approximate a linear curve when the NSSI group (Estimate (SE) =1.96(1.22), z=1.61, p=0.108, 95% CI [−0.43, 4.35]) and SIT group (Estimate (SE) =1.58(1.20), z=1.32, p=0.186, 95% CI [−0.76, 3.93]) were compared with the comparison group. Conflicting emotions in the NSSI group also did not approximate a linear curve when compared to the SIT group (Estimate (SE) =0.38 (0.79), z=0.49, p=0.633, 95% CI [−1.18, 1.93]). The local effect sizes revealed that the variance in the intercept and linear slope increased by 0.16% and reduced by 68.70% respectively for conflicting emotions after adding NSSI and SIT as predictors. 3.3. Post-hoc analysis - duration of time between NSSI or SIT engagement and initiation/termination of distress and negative complex emotions

4. Discussion

Initial changes in distress and negative complex emotions occurred a median of 9.78- (n =34; range 0.58–46.68) and 11.27-h (n =34; range 0.58–46.68), respectively, prior to SIT, and a median of 12.49(n =34; range 0.22–32.60) and 5.54-h (n =35; range 0.23–30.08), respectively, following SIT. In contrast, initial changes in distress and negative complex emotions occurred a median of 19.25- (n =22, range 2.40–85.50) and 16.59-h (n =23; range, 0.86–66. 35), respectively, prior to NSSI and a median of 20.16- (n =22; range 0.64–41.64) and 18.46-h (n =23; range 0.64–55.89), respectively, following NSSI. The trajectory of affect pre- and post-SIT was more variable than that preand post-NSSI. Hence, more participants experienced a reduction or no change in distress and negative complex emotions prior to SIT (distress, n =8; negative complex emotions, n =8) and following SIT (distress, n =9; negative complex emotions, n =5), compared with changes prior to and following NSSI (distress, n =2; negative complex emotions, n =1). As changes in conflicting emotions pre- and post-NSSI and SIT did not differ significantly from the comparison group, the timing of the changes in conflicting emotions were not assessed.

This is the first study to empirically investigate real-time changes in negative complex and conflicting emotions and their relationships with Table 3 Zero order correlations between emotion regulation indices and changes in NCE and CE between the time point prior to and at reported engagement in NSSI, SIT and a random point. Group

NSSI (n = 20) Change in NCE Change in CE SIT (n = 38) Change in NCE Change in CE Comparison (n = 33) Change in NCE Change in CE

3.4. Non-acceptance of negative emotions and multiple emotions prior to NSSI and SIT (Hypothesis 3)

Non-Acceptance of Negative Emotions

Lack of Clarity of Emotions

Lack of Awareness of Emotions

0.70** −0.05

0.39 0.03

−0.11 −0.03

0.46** −0.27

−0.06 0.15

−0.20 0.07

0.21 −2.76

0.183 −0.22

−0.42** 0.21

Note. Negative Complex Emotions (NCE); Conflicting Emotions (CE); Group who engaged in Non-suicidal self-injury (NSSI); Group who engaged in self-injurious thoughts only (SIT); Group who neither engaged in non-suicidal self-injury or selfinjurious thoughts (comparison); *p < 0.050**p < 0.010 (2-tailed).

Higher ratings of trait non-acceptance of negative emotions were significantly associated with an increase in negative complex emotions 107

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

or rumination, as compared with participants reporting NSSI (O’Connor et al., 2012). The current study adds to this finding by revealing that changes in negative complex emotions and distress levels pre- and post-SIT were found to be more variable and fleeting than changes pre- and post-NSSI. The faster termination of distress following SIT suggests that these participants were more effective at regulating their emotional experiences, despite experiencing the same intensity of emotions when engaging in SIT and NSSI. Hence, an extended period of worsening affect prior to and following SIT might be seen as a harbinger of later engagement in NSSI, though further research is required to confirm this hypothesis.

non-acceptance of negative emotions prior to engagement in SIT and NSSI. Four major findings emerged from this study that uses a ‘realworld’ sample of acutely unwell, treatment-seeking youth with BPD, who had minimal prior exposure to BPD treatment. 4.1. Changes in distress and negative complex emotions pre- and post-NSSI and SIT (Hypothesis 1 and 2) First, the number of negative complex emotions experienced prior to and following NSSI and SIT increased and decreased concurrently with self-reported levels of distress. Changes in negative complex emotions fitted a quadratic model prior to and following NSSI, whereas this pattern of response was absent for the comparison group. This finding provides empirical support for qualitative evidence of multiple emotional experiences occurring prior to NSSI (Leibenluft et al., 1987) and for Linehan's theory of 'secondary emotions' for BPD which posits that multiple emotions increase distress and the probability of engaging in maladaptive behaviours such as NSSI (Linehan, 1993; Linehan et al., 2007). In contrast, the pattern of negative complex emotions occurring in the 20- and 10-h before and after SIT did not follow a quadratic curve. These findings might be explained by the high levels of within- and between-person variability in this group. Failure to fit a quadratic curve in the 10-h pre- and post-SIT might also be explained by the high levels of missing data that occurred when reducing the time-frame of affect change. Future studies which have higher adherence rates to the protocol should aim to investigate whether a quadratic curve can be fitted to affect changes over this time-frame or a shorter time-frame (e.g., 5-h pre- and post-SIT). The identical trajectory of negative complex emotions and distress prior to and following SIT and NSSI suggests that for youth with BPD, the experience of an increasing number of emotions is indistinguishable from rising levels of distress. Psychological distress has been defined as a perceived inability to cope effectively causing a change in emotionality, discomfort and intrapersonal and interpersonal harm (Ridner, 2004). The current study adds to our clinical understanding of distress, suggesting it also includes the experience of multiple emotions for youth with BPD. It also encourages treatment that focuses on helping individuals to identify the complexity of their emotional experience and the primary emotion that underlies their distress.

4.4. The relationship between acceptance of emotions and negative complex emotions Fourth, participants who rated themselves as less accepting of their negative emotional responses were more likely to show an increase in the number of negative complex emotions experienced prior to engaging in SIT and NSSI. This accords with prior research, which indicated that improved emotional acceptance is associated with reduced emotion dysregulation and self-harm attempts (e.g., Gratz and Gunderson, 2006). This also provides some preliminary evidence that greater acceptance of emotional experiences might reduce the quantity of emotions experienced and levels of distress, which is the premise of emotion regulation interventions in therapies such as Dialectical Behaviour Therapy (Linehan, 2014). However, further longitudinal research that includes ratings of emotional acceptance is required to confirm this. A significant correlation was found between higher ratings of emotional awareness and increases in the number of negative emotions in the comparison group, suggesting that, in moments of relative non-distress, emotional awareness is important for identifying additional emotions. 4.5. Limitations A limitation to the current study is that participants might have selected more than one emotion because they were unclear or unaware of the emotion they were experiencing. Differentiating emotions is characteristically challenging both for people who engage in self-injury (Bresin, 2014) and for adults with BPD (Suvak et al., 2011; Zaki et al., 2013). If multiple emotions were selected for this reason then lower levels of awareness and clarity should have been significantly correlated with increases in the number of negative complex emotions during distress (prior to NSSI and SIT) but this correlation was not found. In fact, during periods of relative non-distress (for the comparison group) higher levels of awareness were associated with increases in the number of negative emotions. The use of a PANAS item to reflect psychological distress, rather than a separate questionnaire, is also a limitation. It is likely, however, that the inclusion of additional questions in Mobiletype© might have increased participant burden and further reduced adherence rates, which might also limit generalisability of results. Although adherence rates in this study were moderate (51%), they are comparable to other experience sampling studies in undergraduate (38%; Armey et al., 2011) and outpatient populations with schizophrenia (59%; Hartley et al., 2014) and only somewhat lower than in much less acute community samples of adults with BPD (79%; Snir et al., 2015; 72%; Zaki et al., 2013). The lower than desirable adherence rate likely reflects the participants’ high level of clinical acuity and also the difficulty of answering questions during class-time for the 21% of participants who were attending school. This reflects the unavoidable difficulties of conducting research with this population and suggests that extending the window of data collection might not have improved adherence rates. Reliance on trait levels of self-reported non-acceptance of negative emotions also limits the conclusions that might be drawn about how changes in acceptance levels relate to fluctuations in the number of negative emotions

4.2. Changes in conflicting emotions pre- and post-NSSI and SIT (Hypothesis 1 and 2) Second, in contrast with predictions, changes in conflicting emotions did not mirror the changes in self-reported levels of distress and did not significantly increase prior to, and reduce following, SIT or NSSI. This contrasts with Ebner-Priemer and colleagues’ (2008) finding that conflicting emotions were related to the experience of self-reported distress in both healthy controls and adults with BPD. It is possible that the highly symptomatic BPD youth in the current study might have experienced greater difficulty identifying and coordinating conflicting emotions, given their limited treatment experience (in contrast with Ebner-Priemer's study in which 42% of adult BPD participants were undergoing DBT) and lower levels of developmental maturity, which has been found to correspond with poorer emotional clarity (Levine et al., 1997). 4.3. A Post-hoc analysis - duration of time between NSSI or SIT engagement and initiation/termination of distress and negative complex emotions Third, the intensity of distress reported while engaging in SIT and that reported during NSSI were not significantly different. This result is consistent with findings from a retrospective study of 5604 school pupils (15–16 years of age), which revealed that participants who engaged only in SIT did not differ in their levels of anxiety, low mood, 108

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

Chanen, A.M., 2015. Borderline personality disorder in young people: are we there yet? J. Clin. Psychol. 71, 778–791. http://dx.doi.org/10.1002/jclp.22205. Chanen, A.M., Jovev, M., Jackson, H.J., 2007. Adaptive functioning and psychiatric symptoms in adolescents with borderline personality disorder. J. Clin. Psychiatry 68, 297–306. Chanen, A.M., Jackson, H., Cotton, S., Gleeson, J., Davey, C., Betts, J., Reid, S., Thompson, K., McCutcheon, L., 2015. Comparing three forms of early intervention for youth with borderline personality disorder (the MOBY study): study protocol for a randomised controlled trial. Trials 16, 1–10. Chapman, A., Dixon-Gordon, K., 2007. Emotional antecedents and consequences of deliberate self-harm and suicide attempts. Suicide Life-Threat. Behav. 37, 543–552. Chapman, A., Specht, M., Cellucci, T., 2005. Borderline personality disorder and deliberate self-harm: does experiential avoidance play a role? Suicide Life-Threat. Behav. 35, 388–399. Chapman, A., Gratz, K.L., Brown, M.Z., 2006. Solving the puzzle of deliberate self-harm: the experiential avoidance model. Behav. Res. Ther. 44, 371–394. http://dx.doi.org/ 10.1016/j.brat.2005.03.005. Chapman, A., Rosenthal, M.Z., Leung, D.W., 2009. Emotion suppression in borderline personality disorder: an experience sampling study. J. Pers. Disord. 23, 29–47. Choi, B.H., Pos, A.E., Magnusson, M.S., 2015. Emotional change process in resolving self-criticism during experiential treatment of depression. Psychother. Res. 3307, 1–16. http://dx.doi.org/10.1080/10503307.2015.1041433. Conklin, C.Z., Bradley, R., Westen, D., 2006. Affect regulation in borderline personality disorder. J. Nerv. Ment. Dis. 194, 69–77. http://dx.doi.org/10.1097/ 01.nmd.0000198138.41709.4f. Eatough, V., Smith, J.A., Shaw, R., 2008. Women, anger, and aggression. J. Interpers. Violence 23, 1767–1799. http://dx.doi.org/10.1177/0886260508314932. Ebner-Priemer, U.W., Welch, S.S., Grossman, P., Reisch, T., Linehan, M.M., Bohus, M., 2007. Psychophysiological ambulatory assessment of affective dysregulation in borderline personality disorder. Psychiatry Res. 150, 265–275. http://dx.doi.org/ 10.1016/j.psychres.2006.04.014. Ebner-Priemer, U.W., Santangelo, P., Kleindienst, N., Tuerlinckx, F., Oravecz, Z., Bohus, M., Kuppens, P., 2015. Unraveling affective dysregulation in borderline personality aisorder: a theoretical model and empirical evidence. J. Abnorm. Child Psychol. 124, 186–198. Ebner-Priemer, U.W., Kuo, J., Scholtz, W., Kleindienst, N., Rosenthal, Z., Detterer, L., Linehan, M., Bohus, M., 2008. Distress and affective dysregulation in patients with borderline personality disorder: a psychophysiological ambulatory monitoring study. J. Nerv. Ment. Dis. 196, 314–320. http://dx.doi.org/10.1097/ NMD.0b013e31816a493f. First, M., Spitzer, R., Gibbon, M., Williams, J., 2002. Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P) Biometrics Research. New York State Psychiatric Institute, New York. First, M., Gibbon, M., Spitzer, R., Williams, J., Benjamin, L., 1997. Structured Clinical Interview for DSM-IV Axis II Personality Disorders, (SCID-II). American Psychiatric Press, Washington, D.C. Gratz, K.L., Roemer, L., 2004. Multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. J. Psychopathol. Behav. Assess. 26, 41–54. Gratz, K.L., Gunderson, J., 2006. Preliminary data on an acceptance-based emotion regulation group intervention for deliberate self-harm among women with borderline personality disorder. Behav. Ther. 37, 25–35. http://dx.doi.org/10.1016/ j.beth.2005.03.002. Guertin, T., Lloyd-Richardson, E., Spirito, A., Donaldson, D., Boergers, J., 2001. Selfmutilative behavior in adolescents who attempt suicide by overdose. J. Am. Acad. Child Adolesc. Psychiatry 40, 1062–1069. Harter, S., 1983. Children's understanding of multiple emotions: a cognitivedevelopmental approach. In: Overton, W. (Ed.), The Relationship between Social and Cognitive Development. Erlbaum, Hillsdale, NJ, 147–194. Hartley, S., Haddock, G., Vasconcelos e Sa, D., Emsley, R., Barrowclough, C., 2014. An experience sampling study of worry and rumination in psychosis. Psychol. Med. 44, 1605–1614. http://dx.doi.org/10.1017/S0033291713002080. Hox, J., 2010. Quantitative Methodology: Multilevel Analysis: Techniques and Applications 2nd edn. Routledge, Florence, KY, USA. Hulbert, C., Thomas, R., 2010. Predicting self-injury in BPD: an investigation of the experiential avoidance model. J. Pers. Disord. 24, 651–663. http://dx.doi.org/ 10.1521/pedi.2010.24.5.651. Kercher, K., 1992. Assessing subjective well-being in the old-old: the PANAS as a measure of orthogonal dimensions of positive and negative affect. Res. Aging 14, 131–168. http://dx.doi.org/10.1177/0164027592142001. Kleindienst, N., Bohus, M., Ludascher, P., Limberger, M.F., Kuenkele, K., Ebner-Priemer, U.W., Chapman, A.L., Reicherzer, M., Stieglitz, R.-D., Schmahl, C., 2008. Motives for nonsuicidal self-injury among women with borderline personality disorder. J. Nerv. Ment. Dis. 196, 230–236. http://dx.doi.org/10.1097/NMD.0b013e3181663026. Klonsky, E., 2007. The functions of deliberate self-injury: a review of the evidence. Clin. Psychol. Rev. 27, 226–239. http://dx.doi.org/10.1016/j.cpr.2006.08.002. Leibenluft, E., Gardner, D.L., Cowdry, R.W., 1987. The inner experience of the borderline self-mutilator. J. Pers. Disord. 1, 317–325. Levine, D., Marziali, E., Hood, J., 1997. Emotion processing in borderline personality disorders. J. Nerv. Ment. Dis. 185, 240–246. Linehan, M.M., 1993. Cognitive Behavioural Treatment of Borderline Personality Disorder. Guilford Press, New York. Linehan, M.M., 2014. DBT Skills Training Manual 2nd edition. Guilford Press, New York, NY. Linehan, M.M., Bohus, M., Lynch, T.R., 2007. Dialectical behavior therapy for pervasive emotion dysregulation. In: Gross, J. (Ed.), Handbook of Emotion Regulation..

experienced in real-time. To partially address this limitation, future research should include questions relating to non-acceptance of emotions, emotional awareness and clarity in the EMA diary. 4.6. Conclusions The findings from this study elucidate experiential differences between SIT and NSSI, as well as the influence of the quantity of negative emotions and non-acceptance of negative emotions on these distress-related thoughts and behaviours. As participants had not previously undergone evidence-based BPD treatment that might modify their emotion regulation skills, this study offers insights into the naturally occurring experiences of emotion dysregulation prior to NSSI in treatment seeking youth with BPD. It is clinically noteworthy that the majority of participants engaging in SIT experienced a shorter duration of worsening affect and a faster return to baseline compared with participants engaging in NSSI, indicating enhanced emotion regulation skills. This study also found that, prior to NSSI, an ‘avalanche’ of negative emotions coincided with experiences of increasing distress. This suggests that experiences of distress and engaging in NSSI are associated with the number of negative emotions as well as the intensity of emotions (e.g., Andrewes et al., 2016). The finding that higher trait levels of non-acceptance of negative emotional responses are associated with an increased number of negative emotions prior to SIT and NSSI indicates that a failure to accept emotional experiences might trigger an increasing number of negative emotions and distress, as proposed by Linehan (1993, 2007). This provides empirical support for interventions focussed on improving awareness and acceptance of emotions to reduce distress. Funding This work was supported by a National Health and Medical Research Council (NHMRC) Project Grant (GNT0628739). Professor Sue Cotton is supported by an NHMRC Career Development Fellowship (APP1061998). The content is solely the responsibility of the authors and does not necessarily represent official views of the NHMRC. Acknowledgements A special thanks to the patients, families and staff of the HYPE Program at Orygen Youth Health and headspace Sunshine. We would also like to thank Dr Sharnel Perera, Sinn Yuin Chong, Victoria Rayner and Francesca Kuperman for their assistance with data collection. References Andrewes, H.E., Hulbert, C., Cotton, S.M., Betts, J., Chanen, A.M., 2016. Ecological momentary assessment of nonsuicidal self-injury in youth with borderline personality disorder. Personal. Disord. Theory Res. Treat.. http://dx.doi.org/ 10.1037/per0000205. Anestis, M., Silva, C., Lavender, J., Crosby, R., Wonderlich, S., Engel, S., Joiner, T., 2012. Predicting non-suicidal self-injury episodes over a discrete period of time in a sample of women diagnosed with bulimia nervosa: an analysis of self-reported trait and ecological momentary assessment based affective lability and previous suicide attempts. Int. J. Eat. Disord. 45, 808–811. http://dx.doi.org/10.1002/eat.20947. Armey, M.F., 2012. Ecological momentary assessment and intervention in nonsuicidal self-Injury: a novel approach to treatment. J. Cogn. Psychother. Int. Q. 26, 299–318. Armey, M.F., Crowther, J., Miller, I., 2011. Changes in ecological momentary assessment reported affect associated with episodes of nonsuicidal self-injury. Behav. Ther. 42, 579–588. http://dx.doi.org/10.1016/j.beth.2011.01.002. Bresin, K., 2014. Five indices of emotion regulation in participants with a history of nonsuicidal self-injury: a daily diary study. Behav. Ther. 45, 56–66. http:// dx.doi.org/10.1016/j.beth.2013.09.005. Brown, M.Z., Comtois, K.A., Linehan, M.M., 2002. Reasons for suicide attempts and nonsuicidal self-injury in women with borderline personality disorder. J. Abnorm. Psychol. 111, 198–202, doi:10.1037//0021-843X.111.1.198. Carpenter, R.W., Trull, T.J., 2013. Components of emotion dysregulation in borderline personality disorder: a review. Curr. Psychiatry Rep. 15, 335. http://dx.doi.org/ 10.1007/s11920-012-0335-2.

109

Psychiatry Research 252 (2017) 102–110

H.E. Andrewes et al.

69, 1213–1227. http://dx.doi.org/10.1002/jclp.21966. Shiffman, S., Stone, A., Hufford, M., 2008. Ecological momentary assessment. Annu. Rev. Clin. Psychol. 4, 1–32. http://dx.doi.org/10.1146/ annurev.clinpsy.3.022806.091415. Singer, J., Willett, J., 2003. Applied Longitudinal Data Analysis: modeling Change and Event Occurrence. Oxford University Press, London. Snir, A., Berenson, K., Downey, G., 2015. Explicit and inferred motives for nonsuicidal self-injurious acts and urges in borderline and avoidant personality disorders. Personal. Disord. Theory Res. Treat. 6, 267–277. http://dx.doi.org/10.1037/ per0000104. Suvak, M.K., Litz, B.T., Sloan, D.M., Zanarini, M.C., Barrett, L.F., Hofmann, S.G., 2011. Emotional granularity and borderline personality disorder. J. Abnorm. Psychol. 120, 414–426. http://dx.doi.org/10.1037/a0021808.Emotional. Svaldi, J., Dorn, C., Matthies, S., Philipsen, A., 2012. Effects of suppression and acceptance of sadness on the urge for non-suicidal self-injury and self-punishment. Psychiatry Res. 200, 404–416. http://dx.doi.org/10.1016/j.psychres.2012.06.030. Tabachnick, B., Fidell, L., 2013. Using Multivariate Statistics 6th ed. Pearson Publishing, Boston, USA. Vansteelandt, K., Claes, L., Muehlenkamp, J., De Cuyper, K., Lemmens, J., Probst, M., Vanderlinden, J., Pieters, G., 2013. Variability in affective activation predicts nonsuicidal self-injury in eating disorders. Eur. Eat. Disord. Rev. 21, 143–147. http:// dx.doi.org/10.1002/erv.2220. Vinson, T., 2007. Dropping off the edge. Jesuit Social Services/Catholic Social Services Australia, Melbourne. Witre, M., Vallance, D., 1994. A developmental sequence in the comprehension of emotions: intensity, multiple emotions, and valence. Dev. Psychol. 30, 509–514. http://dx.doi.org/10.1037/0012-1649.30.4.509. Yuan, K., Bentler, P.M., 2000. Three likelihood-based methods for mean and covariance structure analysis with nonnormal missing data. Sociol. Methodol. 5, 165–200. Zaki, L., Coifman, K., Rafaeli, E., Berenson, K., Downey, G., 2013. Emotion differentiation as a protective factor against nonsuicidal self-injury in borderline personality disorder. Behav. Ther. 44, 529–540. http://dx.doi.org/10.1016/ j.beth.2013.04.008.

Guilford Press, NY, 581–605. Maas, C.J.M., Hox, J.J., 2004. The influence of violations of assumptions on multilevel parameter estimates and their standard errors. Comput. Stat. Data Anal. 46, 427–440. http://dx.doi.org/10.1016/j.csda.2003.08.006. Mackinnon, A., Jorm, A., Christensen, H., Korten, A., Jacomb, P., Rodgers, B., 1999. A short form of the positive and negative affect Schedule: evaluation of factorial validity and invariance across demographic variables in a community sample. Pers. Individ. Dif. 27, 405–416. Muehlenkamp, J.J., Gutierrez, P.M., 2007. Risk for suicide attempts among adolescents who engage in non-suicidal self-injury. Arch. Suicide Res. 11, 69–82. http:// dx.doi.org/10.1080/13811110600992902. Nock, M., 2009. Why do people hurt themselves? new insights into the nature and functions of self-injury. Curr. Dir. Psychol. Sci. 18, 78–83. http://dx.doi.org/ 10.1111/j.1467-8721.2009.01613.x. O’Connor, R.C., Rasmussen, S., Hawton, K., 2012. Distinguishing adolescents who think about self-harm from those who engage in self-harm. Br. J. Psychiatry J. Ment. Sci. 200, 330–335. http://dx.doi.org/10.1192/bjp.bp.111.097808. Pearson, M., Brewin, C.R., Rhodes, J., McCarron, G., 2008. Frequency and nature of rumination in chronic depression: a preliminary study. Cogn. Behav. Ther. 37, 160–168. http://dx.doi.org/10.1080/16506070801919224. Peugh, J.L., 2010. A practical guide to multilevel modeling. J. Sch. Psychol. 48, 85–112. http://dx.doi.org/10.1016/j.jsp.2009.09.002. Reid, S., Kauer, S., Dudgeon, P., Sanci, L., Shrier, L., Patton, G., 2009. A mobile phone program to track young people's experiences of mood, stress and coping. Soc. Psychiatry Psychiatr. Epidemiol. 44, 501–507. http://dx.doi.org/10.1007/s00127008-0455-5. Reitz, S., Krause-Utz, A., Pogatzki-Zahn, E.M., Ebner-Priemer, U., Bohus, M., Schmahl, C., 2012. Stress Regulation and Incision in Borderline Personality Disorder—A Pilot Study Modeling Cutting Behavior. J. Pers. Disord.. http://dx.doi.org/10.1521/ pedi_2012_26_025. Ridner, S.H., 2004. Psychol. Distress.: Concept Anal.. Selby, E., Franklin, J., Carson-Wong, A., Rizvi, S.L., 2013. Emotional cascades and selfinjury: investigating instability of rumination and negative emotion. J. Clin. Psychol.

110