NSSI (Hasking et al. 2016)

Construct

deliberate damage to the body, without intent to die

Definition: Taylor et al., 2018

Definition: Itzhaky 2015

repetitive

socially unacceptable

Between 1 in 5 and 1 in 8 adolescents and young adults engage in NSSI

Transdiagnostic

can exist in the absence of other psychopathology

often comorbid with other psychopathology, but is not an essential symptom of it

Functions (Taylor et al.)

Two factor model

Intrapersonal

emotion regulation

self punishment

Interpersonal

communicate distress

exert influence over others

punish others

supported in factor analysis

Klonsky et al., 2015

Not like the Nock 4 factor model (used by Itzhaky, 2015)

Prevalence (meta-analysis)

a priori decision on which motives to measure

understanding that many NSSI patients have multiple motives, so not intrapersonal comparisons

forensic and military populations excluded

range of age-groups, and setting (clinical/ non-clinical)

only quantitative data included

Likert type scales were reduced to dichotomous measure - this may disguise the true picture.

frequency is not included. it could be that those who endorse minority function do in fact injure themselves more repeatedly and or severely

STUDIES WITH YOUNGER CHILDREN SHOWED SMALLER ENDORSEMENTS. TAKE A LOOK!

large between-study variability (over and above SE) as a result of

adolescent populations (higher) v others

ISAS measure (higher than FSAM) - double!?

FASM measure (lowest)

most likely because rarely was not considered endorsement

predominantly females

Key findings

intrapersonal more common than interpersonal

emotion regulation most common

Prediction

Itzhaky et al. (2015)

Regression of depression, SC and dependency on NSSI (2 or more incidents in past 12 months)

In the first sample

98% sensitivity

33% specificity

NSSI and SC correlation = .39

In the second sample

90% sensititvity

63% specificity

adding in suicidal ideation improves to 88% specificty

NSSI and SC correlation = .65

Regression on specific functions of NSSI

for 3 functions no significant prediction

suggests that there is overlap in motivations amongst all NSSI users

for positive reinforcement interaction between SC and dependency

if both are high then no NSSI (perhaps issues with self-efficacy)

limitations

almost entirely female

only young adolescents

inpatients with comorbid EDs

high comorbid MHDs

different social factors?

NSSI has been predominantly associated with BN/ purging (See Zelkowitz et al., 2018). This sample was predominantly AN. Nevertheless there was still a strong comorbidity. Questions about generalisability of findings. The high rates in this sample raise similar questions.

The two samples here also show this - the correlation was almost double in the second sample

See Zelkowitz et al., 2018 - meta-analysis

in clinical samples

47 (33,59)

in community samples

36 (25,45)

This is explanation and not prediction (see Fox who cites Yarkoni & Westfall, 2017)

a binary DV. (If a continuous DV is used, better prediction is possible (Fox et al., 2015)

BOTTOM LINE. no evidence for specificity. See Fox et al. 2019 (this is too simplistic - they are already inpatients here!)

Fox et al., 2015

Meta-analysis of 168 predictor variables on NSSI

5078 participants (blend of adult and adolescent)

blend of clinical and community samples (and a minority with past NSSI history)

was this included as a moderator?

central tendency of follow-up was 1-2 years

2/3 used a binary NSSI outcome.

large amount of heterogeneity - therefore CI are most relevant

Once corrected for publication bias, overall risk factor OR = 1.16 (1.10, 1.24)

findings are difficult for most of the theories to account for (small associations with the supposedly big correlates)

moderators

continuous DV measures had better prediction scores (1.5 times more)

harder to predict in general population than in clinical samples. Easiest in past-NSSI group

harder to predict in adolescents than adults (1.5 times)

this may be due to more binary DVs in adolescents

binary IV measures were stronger predictors (1.5 times more)

altogether 52% of variance explained by moderators

Once moderators included and then trimmed for publication bias, very small effects in some populations

limitations

see moderators

this may not apply to short-term prediction, which is the norm in clinical settings

many binary IV measures of past NSSI allowed one instance to be included

no standardised definition of the severity of self-harm necessary to be considered NSSI

studies look at risk factors for continued NSSI, but not for the onset

key

look out for suicide ideation - strong association

Fox et al., 2019

Machine learning algorithm to consider complex relationships (interactions, non-linear) between the many predictors (e.g. Fox et al., 2015).

random forest model

short-term follow-ups at 3, 14, and 28 days

1021 participants recruited from online psychopathology (Which?)/ suicide forum

does this generalise to community samples?

88% with lifetime NSSI history

96% with suicide plans history

68% suicide attempt

past year suicide attempt - or 2 cutting episodes in past 2 weeks - or frequent suicide ideation in past 2 weeks

moderator

not very generalisable - this does not predict onset

specific focus on cutting

39 predictor variables

6 demographic

12 cognitive/ emotional

7 past (intent) NSSI

11 past suicide

4 (2*2) test of stimuli (positive/ death)(explicit/ implicit)

compared univariate prediction with multivariate prediction with forest plot

univariate

chance sensitivity and specificty

multivariate

a bit better than chance

forest plot

over 90% specificity, over 75% sensitivity

within the context of the model, the most powerful predictor was number of self-cutting incidents in past month. BUT PREDICTION WAS VIRTUALY IDENTICAL WHEN THIS WAS REMOVED.

NOTE. this does not represent a linear relationship between the IV and DV

suggests no singular recipe for predicting the DV

high sufficiency, low necessity pattern = complex adaptive system. thought to be widespread in psychopathology.

CONCLUSION:

these models are the way forward

findings support that there is no one causal pathway. a complex adaptive system

Preschoolers

Luby et al., 2019

on the up (50% increase over the past 20 years)

one community sample reported a prevalence > 7%

children recruited for a depression intervention research

the healthy controls are therefore not typical healthy controls

children on medication excluded

64% male, 72.6% Caucasian, mean age 5.15

the sample was N (314), 26 subclinical and 288 with MDD

this messes up comaprisons

19.4% suicidal ideation

5.4% of sample also reported NSSI

(So NSSI = 21.3% of sample)

1.6% of sample had attempted

15.9% NSSI

anyone with NSSI and SI was upgraded

this messes up NSSI v other group matches

64.6% None

Findings

very small effects for

increased depression severity

(remember all were depressed or almost depressed in the whole sample)

higher irritability/ temper

higher externalising symptoms

perhaps family history of bipolar (missed signficance)

more violent life events

(look at CI - could be quite a large effect)

NO DIFFERENCE IN IMPULSIVITY

limitations

all treatment seeking

largely caucasian

relies on parent report

all depressed

Dervic & Oquendo, 2019

even though population prevalence is low, clinical professionals will meet them

three points on Luby et al'. study

1. higher prevalence of SI an SB than previously reported

MDD sample

parent report

2. no association with parental mood disorders, unlike previous research

MDD sample

parent report

NSSI is a big predictor for suicide. Red flag

Preoccupation with death important to distinguish between NSSI only children and NSSI + SI

interventions include cognitive targeting guilt and medication when necessary

Treatment

will differ in line with NSSI function (Taylor et al., 2018)

perhaps target perfectionism (Egan) / SC (Itzkowitz)

no evidence-based successful treatment (perhaps due to lack of targeting underlying processes) - Fox et al., 2015

NSSI Theories (general)

Social cognitive theory (Bandura, 1997; Hasking et al., 2016)

outcome expectancies: the expected desirability/ outcome of the behaviour (based on direct or observed past experiences)

Theory

strong association between positive outcome expectancies and indirect harmful behaviours (eating, alcohol, smoking)

maintaining or manipulating emotional states is a central function of NSSI (expecting to achieve that emotional outcome is thus fundamental)

Findings

self-efficacy expectancies: the expectation that one is capable of successfully implementing a chosen behaviour

general self-efficacy

the ability to handle situations as they arise. Stressful situations, or socially challenging environments may be especially relevant to NSSI.

trait-specific self-efficacy

central to relapse, and prevention of subsequent episodes once a behaviour takes hold

a more powerful predictor of behaviour

JOINER, 1999

suicide requires self-efficacy for implementation

Why NSSI and not drugs or some other escape?

Cognitive-Emotional model of NSSI (Hasking et al., 2016). Emphasis on points 4 and 5 (beliefs about NSSI)

Theory

1. High emotional reactivity

but why/ moderators/ mediators

2. Poor emotional regulation abilities

see above for models/ competencies

3. Negative self-schema: the way one views oneself (mediates the transition between negative emotional episodes and accumulated/ psychopathological negativity).

4. outcome expectancy: the belief that NSSI will provide relief

5. self-efficacy: belief in the anility to implement the self-injury

Findings

association between outcome expectancies, self-efficacy and NSSI

4 models in Itzkowitz 2015

affect regulation (see intrapersonal below in detail

environmental

synonymous with interpersonal (see Klonsky et al., 2015; Taylor et al., 2018)

self-punishment

relates to childhood emotional abuse

relates to perfectionism and self criticism

NOTE: does this relate to the affect regulation and environmental models, or is it discreet? Klonsky et al. 2015 load this factor onto the intrapersonal factor, suggesting this has to do with affect regulation

psychodynamic

NSSI intrapersonal

To avoid difficult emotions (Emotional regulation deficits)

5 emotion regulation models

Process model (Gross, 1998)

5 stages in the process of emotional arousal

Antecedent (pre-arousal)

situation selection

situation modification

attentional deployment

this can include attentional biases

cognitive change

response-focused

emotional suppression

cognitive reappraisal is negatively associated with distress, emotional suppression can exacerbate it

RESEARCH

NSSI patients report emotional distress

CR is negatively associated with NSSI

MY BOTTOM LINE

This is a cognitive (and behavioural) model emphasizing the importance of the correct perspective and interpretation of events. It makes sense that experiencing strong negative feelings results in distress and a search for means to reduce that stress.

Emotional cascade model (Selby & Joiner, 2009)

Emotionally challenging moments can be tolerated, but repetitive rumination and re-experiencing of these moments can cause great distress.

Some complication with distinction between brooding and reflection (what it means + relationship with NSSI)

RESEARCH

many studies have found an association between rumination and NSSI

difficulties in emotion regulation model (Gratz & Roemer, 2004)

Acceptance of emotion - two keys:

awareness and understanding of emotions

acceptance of emotions

Subtopic

Choosing a behavioural response: two keys

impulse control

flexibility/ situational appropriate strategy

RESEARCH

mindfulness moderates relationship between depression and NSSI

This is over and above the effects of mindfulness on becoming depressed in the first place

MY BOTTOM LINE:

There's nothing wrong with aroused emotions and response-focused strategies if behavioural suppression is accompanied by the acceptance of powerful emotions and their accompanied physiological arousal. If difficult emotions don't result in distress, it is logical that dramatic actions such as NSSI are less necessary.

Experiential avoidance model (Chapman et al., 2006)

the need to avoid experiencing difficult emotions

NSSI focuses attention away from the emotions

NSSI can become a classically conditioned response to difficulty

RESEARCH

higher avoidance amongst NSSI than controls (both clinical BPD and community)

especially high amongst individuals with comorbid NSSI and ED

MY BOTTOM LINE: This has much in common with the previous model. The root of the problem is the inability to experience challenging emotions. This model adds expressly continued by explaining how NSSI is recruited to solve the problem, and how it can become a habit.

emotional reactivity (Hasking et al. 2016)

more easily aroused + greater aroused intensity

most naturally feeds into mindfulness based models

RESEARCH

emotional reactivity is associated with NSSI (but not hen physiological laboratory methods are used)

low positive reactivity has a stronger association with NSSI than negative emotion reactivity (Gratz et al., 2006 - Is Allen & Hooley, 2015 related)?

NSSI-Perfectionism

Theory

Subtopic

Findings (Egan et al., 2011)

socially prescribed perfectionism predicts self-harm

My bottom line: This effect is probably associated with the ER models described above. Is perfectionism the source of emotional cascades and avoidance? Or perhaps perfectionism is itself an avoidance behaviour? Fascinating question!

NSSI - SC/ dependency

Impulsivity

Theory

Findings

Self report suggests that yes (more self-critical?)

Behavioural/ computerised measures suggest that no