Received: 29 October 2018 Revised: 18 April 2019 Accepted: 25 April 2019

DOI: 10.1002/cpp.2372

COMPREHENSIVE REVIEW

Self‐compassion, self‐forgiveness, suicidal ideation, and

self‐harm: A systematic review

Seonaid Cleare | Andrew Gumley | Rory C. O'Connor

Suicidal Behaviour Research Laboratory,

Institute of Health and Wellbeing, University Abstract
of Glasgow, Glasgow, UK Self‐compassion has been implicated in the aetiology and course of mental health
Correspondence with evidence suggesting an association between greater self‐compassion and lower
Seonaid Cleare, Suicidal Behaviour Research emotional distress. However, our understanding of the nature and extent of the rela-
Laboratory, Institute of Health and Wellbeing,
University of Glasgow, 1055 Great Western tionship between self‐compassion and self‐harm (self‐injury regardless of suicidal
Road, Glasgow G12 0XH, UK. intent) or suicidal ideation remains unclear. This review, therefore, aimed to critically
Email: [email protected]
evaluate the extant literature investigating this relationship. To do so, a systematic
search, including terms synonymous with self‐compassion, was conducted on three
main psychological and medical databases (Web of Science, PsycINFO, and Medline).
Only studies investigating self‐compassion or self‐forgiveness and self‐harm or suicidal
ideation were found to be relevant to the review. Eighteen studies were included in
the final narrative synthesis. Heterogeneity of studies was high, and the majority of
studies were quantitative and cross‐sectional (n = 16) in design. All studies reported
significant associations between higher levels of self‐forgiveness or self‐compassion
and lower levels of self‐harm or suicidal ideation. Several studies suggested that
self‐compassion or self‐forgiveness may weaken the relationship between negative
life events and self‐harm. In conclusion, this review highlights the potential importance
of self‐compassion in the aetiology of suicidal thoughts and self‐harm. We discuss the
clinical and research implications.

K E Y W OR D S

self‐compassion, self‐forgiveness, self‐harm, suicidal ideation, suicide attempt

1 | I N T RO D U CT I O N Despite our understanding of risk factors, there are many gaps in

our knowledge; indeed, we are unable to accurately predict those
Suicide is a major public health concern, with approximately 804,000 who are at risk of suicide (Franklin et al., 2017). To date, the most con-
people dying by suicide annually (World Health Organization [WHO], sistent predictor of a suicide attempt is having made a previous suicide
2014). It is well established that suicidal thoughts and behaviours attempt (Arensman, Griffin, & Corcoran, 2016). Having engaged in
result from an interplay of biological, psychological, clinical, cultural, nonsuicidal self‐injury (NSSI) also increases an individual's risk of future
and social factors (O'Connor & Nock, 2014), and much of the research suicidal behaviour (Chan et al., 2016; Kiekens et al., 2018; Ribeiro et al.,
to date has sought to identify and understand how specific markers 2016), with around 50% of people who die by suicide having self‐
contribute to an individual's risk of suicide. Psychological risk markers harmed previously (Foster, Gillespie, McClelland, & Patterson, 1999).
such as self‐criticism, shame, perfectionism, isolation, entrapment, and For the present purposes, self‐harm is defined as “self‐injury or self‐
perceived burdensomeness are repeatedly implicated in suicide risk poisoning irrespective of the apparent purpose of the act” (National
(O'Connor & Nock, 2014). Institute for Health and Care Excellence [NICE], 2004, 2011).

Clin Psychol Psychother. 2019;1–20. wileyonlinelibrary.com/journal/cpp © 2019 John Wiley & Sons, Ltd. 1
2 CLEARE ET AL.

The inability to identify those most at risk of self‐harm and suicide

is in part because previous research has not been guided sufficiently
Key practitioner message
by theoretical models. The Integrated Motivational–Volitional (IMV)
model of suicidal behaviour is a tripartite (premotivational, motiva- • Higher self‐compassion and self‐forgiveness are
tional, and volitional phases) diathesis–stress framework that incor- associated with lower levels of self‐harm and suicidal
porates major components from psychopathology, suicidal behaviour ideation.
research, and health psychology literature to delineate the final com- • Research into this area is limited, study heterogeneity
mon pathway to ideation and enactment of self‐harm and suicidal was high, and designs tended to be cross‐sectional.
behaviour (O'Connor, Cleare, Eschle, Wetherall, & Kirtley, 2016; More prospective studies are needed.
O'Connor & Kirtley, 2018; O'Connor, 2011).
• There are some indications that self‐compassion and self‐
The IMV maps out a detailed path from background context (e.g.,
forgiveness may alter the relationship between negative
deprivation, genetics, and negative life events) in which self‐harm ide-
life events and self‐harm.
ation may develop. The motivational phase highlights factors that may
facilitate the transition from defeat to entrapment (threat to self‐
moderators; e.g., rumination and problem solving) and entrapment to personal shortcomings rather than being critical of them; having a
self‐harm ideation (motivational moderators; e.g., resilience and social mindful awareness of thoughts, emotions, and experiences that are
support). The volitional phase outlines factors that influence the likeli- emotionally painful; and actively adopting a warm and supportive
hood that someone engages in self‐harm (volitional moderators; e.g., response to these experiences rather than judging the self harshly
having access to means and reduced sensitivity to pain). There has for these events. Additionally, it entails acknowledging that failure is
been a growing body of evidence supporting these relationships something that everyone experiences rather than feeling isolated by
(Johnson, Wood, Gooding, Taylor, & Tarrier, 2011; O'Connor, 2003; experiences (Neff, 2003a, b; Neff, 2016).
O'Connor, Smyth, Ferguson, Ryan, & Williams, 2013; Rasmussen Neff describes self‐compassion as a balancing of six integrally con-
et al., 2010). nected elements:
The IMV highlights the complex interplay between risk and poten-
tial protective factors (O'Connor & Nock, 2014). These protective fac- self‐kindness – extending kindness and understanding to

tors may be crucial in understanding and protecting against risk of oneself in instances of perceived inadequacy or suffering
self‐harm by, for example, buffering the impact of stressful life events rather than harsh judgement and self‐criticism, common
(O'Connor & Nock, 2014). Self‐compassion is one such protective fac- humanity – seeing one's experiences as part of the

tor that has received considerable attention in the aetiology of mental larger human experience rather than seeing them as
and physical health. The role of self‐compassion within the IMV model separating and isolating, and mindfulness – holding
is not yet known. However, the affiliative nature of compassion may one's painful thoughts and feelings in balanced
make it effective in reducing social threat‐based emotions, such as awareness rather than over‐identifying with them in an
shame and defeat, thereby suggesting that self‐compassion is a mod- exaggerated manner. (Neff & Lamb, 2009, p. 864)
erator within the motivational phase or it may operate throughout Each component reinforces another (Neff, 2003a; Barnard & Curry,
the pathway. 2011); for instance, feeling connected to others reduces feelings of
isolation, leading to individuals feeling more positive about themselves.
1.1 | What is self‐compassion?

Compassion is a multifaceted construct, which develops within a 1.2 | Measuring self‐compassion

secure attachment framework (MacBeth & Gumley, 2012), and has
been conceptualized in various ways (see Gilbert, 2017 and Kirby, The most widely used measure of self‐compassion is the Self‐
2017 for a review and discussion of the different definitions). Compassion Scale (SCS; Neff, 2003b). Neff (2003a, b, 2016) described
One of the more frequently used definitions of compassion is self‐compassion as requiring an interaction between the positive and
based on the Buddhist conceptualization of compassion as a motiva- negative components of compassion and, consequently, developed
tion to prevent suffering of self and others: the SCS to assess compassion as per her definition (Neff, 2003b).
There has been considerable debate regarding the validity of the
Being sensitive to the suffering of self and others with a
SCS as a measure of self‐compassion. In particular, concerns have
deep commitment to try to prevent and relieve it.
been expressed that by including “negative” components of compas-
(Gilbert & Choden, 2013, p. xxv)
sion, the SCS measures self‐criticism, rumination, and social isolation
Self‐compassion, then, is more than the absence of self‐criticism. (MacBeth & Gumley, 2012; Muris, 2016) and that using the total score
Rather, it is a process in which individuals have the intention and moti- might lead to an overestimation of the relationship with symptoms of
vation to adopt and apply a compassionate mindset to themselves psychopathology as the negative components are more strongly asso-
(Jazaieri et al., 2014). For instance, self‐compassion entails accepting ciated with psychopathology than the positive components (Muris &
CLEARE ET AL. 3

Petrocchi, 2017). In light of these concerns, the psychometric proper- anxiety and well‐being (Sommers‐Spijkerman, Trompetter, Schreurs,
ties of the SCS have been extensively investigated. Taken as a whole, & Bohlmeijer, 2018) through increasing positive affect, which subse-
research has yielded support for a model in which the interrelated sub- quently reduced levels of depressive symptoms. Compassion‐focussed
scales are encompassed by an overarching self‐compassion factor. therapy also reduced self‐criticism, which in turn reduced symptoms
This is consistent with Neff's assertion that both the SCS subscale of anxiety. Indeed, studies using functional magnetic resonance
scores and overall self‐compassion score are valid (Cleare, Gumley, imaging have shown that areas of the brain associated with affect
Cleare, & O'Connor, 2018; Neff et al., 2019; Neff, Whittaker, & Karl, regulation, reward, and affiliation activate in response to compassion
2017; Tóth‐Király, Bőthe, & Gábor, 2017). Several alternative models (Colonnello, Petrocchi, & Heinrichs, 2017; Leiberg et al., 2011; Lutz,
for the SCS have also been proposed, including a two‐factor model Slagter, Dunne, & Davidson, 2008). Subsequently, self‐compassion
based on the SCS scoring methods (i.e., self‐coldness [reverse scored may have a role in ameliorating the impact of personality traits often
items]) and self‐compassion (Gilbert, McEwan, Matos, & Rivis, 2011); implicated in self‐harm such as self‐criticism and perfectionism
however, this model has not been supported by subsequent analyses (O'Connor, 2011; O'Connor & Nock, 2014).
(Cleare et al., 2018; Neff et al., 2019). One of the challenges facing self‐compassion researchers is the
range of terms used interchangeably with self‐compassion. Barnard
and Curry (2011) discuss the differences between many related terms
1.3 | Self‐compassion and well‐being (i.e., self‐esteem and empathy) and self‐compassion. Since their review,
however, there has been an increase in self‐forgiveness research, which
Increasingly, self‐compassion has been shown to be associated with is important to consider as a possible component of self‐compassion.
physical (r = .23 to .28; Hall, Row, Wuensch, & Godley, 2013) and psy- However, it should be noted that self‐compassion requires the indi-
chological well‐being (positive affect r = .36; anxiety r = −.58; and vidual to have feelings of warmth towards the recipient (Gilbert,
depression r = −.46; see Barnard & Curry, 2011 for review), including 2017), whereas this is not necessary in forgiveness.
reduced emotional burnout and shame (r = −.6). Using meta‐analytic
techniques, MacBeth and Gumley (2012) found higher self‐compassion 1.4 | What is self‐forgiveness?
was associated with lower levels of depression, anxiety, and stress
(r = −.54, 95% CI [−0.57, −0.51]). Both the review and meta‐analysis Self‐forgiveness can be conceptualized as an emotion regulation pro-
emphasize that the majority of studies were cross‐sectional and the cess, which begins when an individual accepts responsibility for their
direction of the relationship is unknown, although the literature sug- actions, feels remorse and guilt, and begins to release self‐directed neg-
gests that the absence self‐compassion is more likely to lead to emo- ativity and begins to heal themselves (Enright, 1996; Wohl, DeShea, &
tional distress rather than vice versa. Wahkinney, 2008). It has recently been defined as follows:
Psychological intervention studies found participants who engaged
Self‐forgiveness … is a deliberate, volitional process
with repeated compassionate meditations reported reductions in neg-
initiated in response to one's own negative feelings in
ative emotions, including feelings of shame and self‐criticism (Gilbert &
the context of a personally acknowledged self‐instigated
Procter, 2006), lower symptoms of illness, and higher social support and
wrong, that results in ready accountability for said
higher life purpose (Fredrickson, Cohn, Coffey, Pek, & Finkel, 2008).
wrong and a fundamental, constructive shift in one's
Interventions have been found to be effective across a range of
relationship to, reconciliation with, and acceptance of
populations, including student (Smeets, Neff, Alberts, & Peters, 2014),
the self through human connectedness and commitment
adolescent (Bluth & Eisenlohr‐Moul, 2017; Mcgehee, 2010), and clinical
to change. (Webb, Bumgarner, Conway‐Williams, Dangel,
populations including borderline personality disorder (Krawitz, 2012),
& Hall, 2017, p217)
populations with depression (Gilbert & Procter, 2006), schizophrenia
spectrum disorders with psychotic features (Braehler et al., 2013) and This definition echoes aspects of self‐compassion. Specifically, the
forensic mental health inpatient populations (Laithwaite et al., 2009). motivation to accept the self, including flaws, whilst recognizing
Even single‐session compassion inductions have been shown to the need to make changes or take reparative action has parallels
reduce negative emotions (Arimitsu & Hofmann, 2017), raise mood, with self‐kindness. The emphasis on feeling connected to others as
and increase positivity towards others (Hutcherson, Seppala, & a mechanism to support self‐acceptance is akin to common human-
Gross, 2008). ity. In these instances, a mindful attitude rather than rumination
Despite the association between self‐compassion and psychologi- may help reconciliation with the self. Indeed, Hirsch, Webb, and
cal well‐being, the nature of the relationship between self‐compassion Jeglic (2012) found that self‐forgiveness moderated the relationship
and suicidal ideation or self‐harm is unclear. between internally directed anger and suicidal behaviour even when
Through adopting a compassionate stance to themselves, self‐ external anger was included in the model. Previous research has
compassion may help individuals to tolerate difficult emotions (Gilbert, identified expressions of internally directed anger in suicide notes:
2017; Klimecki, Leiberg, Ricard, & Singer, 2014; Leiberg, Klimecki, & For example, O'Connor, Sheehy, and O'Connor (1999) found that
Singer, 2011). A recent study of self‐help compassion‐focussed ther- 64.3% of note writers who had attempted suicide previously expressed
apy showed that self‐compassion mediated the relationship between self‐directed anger.
4 CLEARE ET AL.

In summary, self‐compassion has associations with other areas of self injur OR self‐harm OR self harm. We used the truncation symbol
mental well‐being and may be an important factor in buffering against (*) to find any different endings to the terms. See Figure 1 for details
suicidality. Consequently, it is important to determine the nature and of the search strategy.
extent of the relationship between self‐compassion and self‐harm, sui-
cide attempts, or ideation. To this end, this systematic review aimed to
critically evaluate the extant research that has investigated the rela-
2.2 | Inclusion and exclusion criteria
tionship between self‐compassion/self‐forgiveness and self‐harm and
To be eligible for inclusion, studies had to (a) assess self‐compassion or
suicidal ideation.
related term; (b) assess self‐harm (with or without suicidal intent) or
suicidal ideation; and (c) record the relationship between self‐
compassion (or related term) and self‐harm or suicidal ideation. We
2 | METHODS
included all ages and participant groups. The reference lists of all the
included papers were hand‐searched. Decisions around inclusion were
2.1 | Search strategy made by the first author in the first instance, with verification from the
second and third authors.
We searched the following relevant databases: Web of Science,
EBSCO Host (Medical and Psychology related resources), PubMed,
CINAHL, and PsycINFO for relevant empirical studies published up 2.3 | Data extraction
to August 2018 with no date limiters used. Searches were constrained
to papers published in peer‐reviewed journals and in English. Demographic characteristics, study design, and assessment of suicidal
The following search terms were employed: self‐compassion or ideation or self‐harm, self‐compassion, or self‐forgiveness were
self compassion OR self‐ empath OR self empath OR self‐forgiv OR self extracted along with the main findings. A quality assessment frame-
forgiv OR self‐car OR self car, OR self sooth OR self‐sooth OR self‐ work based on O'Connor, Ferguson, Green, O'Carroll, and O'Connor
sympath OR self sympath OR self‐warmth OR self warmth OR self‐ (2016) was used to assess study rigour. This scale has nine areas for
kindness OR self kindness OR mutuality; AND suicid OR self‐injur OR consideration (e.g., study design and statistical power/considerations;

FIGURE 1 Procedure for identifying,

screening, and determining the eligibility of
studies for inclusion in the review
CLEARE ET AL. 5

sample details, comparison group, and compassion construct assess- were the only group to report calculations for statistical power. Only
ment) allowing calculation for an overall score for the study ranging seven studies controlled for confounding variables during analysis.
from 0 to 13. For example, a score of “0” is assigned to cross‐sectional
studies, case‐controlled studies are assigned a score of “1,” and pro-
spective studies receive a “2.” In terms of study design, studies were 3.3 | Sample characteristics
also assessed on measures they used (i.e., single items or nonvalidated
The combined sample size was 4,345 participants, with a mean age of
scales scored “0”; validated scales or interviews scored “2”) and
20.9 years old (range = 11–66 years old); 58.6% (n = 2,547) of partic-
whether they included a comparison group. This allows heteroge-
ipants were female. Five studies were conducted in North America
neous research designs to be compared with continuity. As this
(Chang et al., 2017; Gregory et al., 2017; Hayes et al., 2016; Rabon,
framework was not applicable for assessing qualitative studies, we
Sirois, & Hirsch, 2018; Tanaka et al., 2011) and were the only studies
adapted and applied the Critical Appraisal Skills Programme (Critical
to detail ethnicity; three of the samples were predominantly White
Appraisal Skills Programme [CASP], 2017) guidelines to assess appro-
(59–89%) and female (67.9–100% female). Tanaka et al.'s (2011) sam-
priateness of the study design, data collection, and analysis.
ple reported diverse ethnic backgrounds (27% White, 31.3% Black,
and 27.8% dual/multiple ethnicity). Two studies were conducted in
3 | RESULTS China (Jiang et al., 2016; Jiang, You, Zheng, et al., 2017; Jiang, You,
Ren, et al., 2017) and two in Europe (Collett et al., 2016; Xavier
Eighteen papers were included in the review (see Figure 1). Eleven et al., 2016). Collett et al. (2016) carried out a case‐controlled study,
studies addressed self‐compassion (eight cross‐sectional, two longitu- comparing a clinical population (experiencing persecutory delusions
dinal, and one qualitative), and seven addressed self‐forgiveness (all n = 21) with a group with no history of any mental health problems
cross‐sectional). No other synonyms of self‐compassion were eligible. (controls; n = 21). The groups were matched for age and gender (clin-
Where possible, we have reported the effect sizes for correlations (r ical age range = 21–66, m = 45.6 years old; control age range = 22–61,
values). m = 41.9 years old).
Studies reported a range of outcomes, including suicidal behaviours
(combined suicidal ideation and attempts; self‐compassion n = 2, self‐
forgiveness n = 4), NSSI (self‐compassion n = 4, self‐forgiveness 3.4 | Assessment of self‐compassion
n = 1), suicidal ideation (self‐compassion n = 1, self‐forgiveness n = 1),
suicide attempts (self‐compassion n = 1), self‐harm (self‐compassion The SCS (Neff, 2003) was the most frequently used measure; three

n = 1), and multiple aspects of self‐harm (self‐compassion n = 1, studies reported subscale scores and six the total score. Two studies

self‐forgiveness n = 1). The final study was qualitative and used Inter- (Hayes et al., 2016; Rabon et al., 2018) used the 12‐item SCS short

pretive Phenomenological Analysis to assess the self‐compassion in form (Raes et al., 2011). The SCS short form includes two items from

blog posts related to self‐harm. each of the original subscales. In addition to the SCS, Gregory et al.
(2017) measured state self‐compassion (participants rated how
trusting, loving, grateful, and joyful they were feeling) before and after
3.1 | Quantitative studies of self‐compassion
a values affirmation task.

Ten studies were included in this section (see Table 1 for details);
however, two studies (Jiang, You, Zheng, & Lin, 2017; Jiang,
3.5 | Assessment of self‐harm and self‐harm ideation
You, Ren, et al., 2017) appear to report the same study. To avoid
duplication, the sample characteristics from the brief report Four studies used a single item to assess self‐harm or ideation (lifetime
(Jiang, You, Zheng, et al., 2017) are not included, although the history: Gregory et al., 2017; last 12 months: Jiang, You, Zheng, et al.,
findings from both are discussed as they report on different aspects 2017; Jiang, You, Ren, et al., 2017; Tanaka et al., 2011). Although Hayes
of self‐compassion. One study (Collett et al., 2016) was conducted in et al. (2016) recorded lifetime suicidal ideation, suicide attempts, and
a clinical population; four studies were carried out with adolescents NSSI, they reported a dichotomized score indicating the presence or
and four recruited university students. absence of suicidal ideation or self‐harm.
The remaining studies assessed a variety of outcomes, including sui-
3.2 | Quality assessment cidal ideation (Beck Scale for Suicidal Ideation; Beck & Steer, 1991 in
Collett et al., 2016) and self‐harm (Risk‐taking and Self‐harm Inventory
Methodology quality assessment scores (see Table 1 for details) ranged for Adolescents Portuguese; Xavier et al., 2013 in Xavier et al., 2016).
from 2 to 6 (low/medium–high). The majority of studies scored low Two studies (Chang et al., 2017; Rabon et al., 2018) assessed mixed sui-
for their design; six studies were cross‐sectional, and four made no cidal behaviours (Suicidal Behaviours Questionnaire‐revised [SBQ‐R];
attempt to include homogenous groups. Only three studies (Collett Osman et al., 2001). Jiang et al. (2016) assessed the frequency of NSSI
et al., 2016; Gregory et al., 2017; Xavier et al., 2016) used validated methods used in the preceding 12 months with responses on a Likert‐
measures, and all studies used self‐report measures. Collett et al. (2016) type scale ranging from 1 (never) to 7 (almost every day).
 6

TABLE 1 Self‐compassion quantitative studies

Study, Analysis

Measures Key findings Covariates

country, quality
Outcome
assessment (QA)
score Sample Study design Self‐compassion Measure
Chang et al. Students. N = 331 Cross‐sectional; SCS (Neff, Combined suicidal ideation SCS subscales significantly associated Correlations multiple
(2017), USA (F = 225, 67.9%) observational 2003a, b) and suicide attempts with suicidal behaviours (r = .2 to .26) mediation models
QA = 3 in expected directions. (depressive symptoms
Mean age: 21.5 SC potential mediator of NLE last And SB) to assess effect
12 months and SBQ‐R score. of each compassion
Range: 18–58 SBQ‐R (Osman, Bagge, NLE negatively related to common component.
European Gutierrez, Konick, humanity (B = −.11), which in turn
American = 88.8%, Kooper & Barrios, was negatively related to suicidal
African American = 6%; 2001) behaviours (B = −.13). The full model Gender Used p < .10
Asian American = 3.3%; involving NLE and SC facets, controlling significance
Latino = 1.8% for sex, accounted for a small (f 2 = .16)
but significant (13.7%) of variance in
suicidal behaviours, F(7, 323) = 7.18,
p < .001.
Collett, Pugh, Clinical (persecutory Cross‐sectional; SCS (Neff, Suicidal ideation Clinical group lower self‐compassion and Correlations
Waite, and delusions) vs. Case controlled 2003a, b) higher depression than C group (p < .05). Mann–Whitney U‐
Freeman controls (C)
(2016), UK N = 42; 21 clinical, Clinical group recruited BSSI (Beck & Steer, 1991) Self‐compassion negatively correlated Tests Cohen D
QA = 6 21 C Groups matched clinical service; data with suicidal ideation (r = −.64; p = .002) calculated.
age/gender collected interview and measures of self‐cognitions. None
with clinician. C group
Mean age: 45.6, 41.9,
from participant
respectively
pool online.
Range: 21–66
Ethnicity not reported
Gregory, Glazer, Students N = 64; Cross‐sectional; SCS (Neff, Self‐harm SH, M (SD) = 2.40 (0.57), lower trait than Correlations T‐tests
and Berenson all female. experimental 2003a, b) C, M (SD) = 3.25 (0.63), t(62) = −5.68,
(2017), USA SH = 32; p < .001, d = −1.44, and state,
QA = 6 C = 32 State Item from the SNAP‐2 F (1, 60) = −6.69, p = .012, d = −0.66 M Regressions
self‐compassion (Clark, 2003; item 174) (SD) = 3.08 (0.89), than C, M (SD) = 3.60
trusting, loving, assessed repeated (0.84) self‐compassion.
grateful, joyful engagement deliberate
Mean age: 19.4 Post VA: MANCOVA 2×s Design
(not at physical self‐injury.
Range: 18–22 Self‐compassion increased in both SH, M VAS Joyful Trait
all–extremely)
White = 89.1% (SD) = 3.52 (0.70), versus M, (SD) = 2.64 Compassion
(0.85), and C group, M (SD) = 3.77 (0.92),
versus M, (SD) = 3.44 (0.75), than neutral
condition, M (SD) = 3.04 (0.89). SH group
pain endurance reduced to level of C.
CLEARE

(Continues)
ET AL.
TABLE 1 (Continued)
CLEARE

Study, Analysis

Measures Key findings Covariates

ET AL.

country, quality
Outcome
assessment (QA)
score Sample Study design Self‐compassion Measure
Values affirmation produce the greatest
gains in state self‐compassion among
individuals with low in trait self‐compassion.

Hayes, Lockard, Students registered with Cross‐sectional; SCS‐sf (Raes, Suicidal ideation, suicide Factor analysis of SCS‐sf; differences Correlations ANOVAs
Janis, and mental health services, observational Pommier, Neff, attempts, NSSI between groups for total scores reported. None
Locke (2016), 1,609 ( f = 1,110; 69%, & Van Gucht, 2011) ANOVAs conducted C; SI, SA; NSSI
USA
m = 499; 31%) Lifetime frequency.
QA = 3
Mean age: 22.74 Dichotomised score used.
Range: 18–63 (85%
under 25 years old)
European American/
White = 59%
African American/
Black = 13%
Hispanic/Latino/a = 13%
Asian American = 8%
Multiracial = 4%
Other = 2%
Jiang et al. Adolescents 525 Longitudinal SCS (Neff, NSSI in 12m. Time 1: Correlations
(2016), China ( f = 225, 43%) 2003a, b) 152 (29%) engaged in NSSI, 69 (29%) Regressions
QA = 4 1 method, 83 (54.6%) multiple methods. Correlations: Living
Self‐compassion negatively correlated arrangements
Mean age: 12.97 NSSI methods listed with NSSI (r = −.3) and being bullied Parent's education/
frequency scale (Never– (r = −.27; both p < .001) occupation
almost every day)
Time 2: Regressions: T1 NSSI,
Range: 11–16 137 (26.1%) NSSI, 60 (44.1%) 1 method, Bullying, Gender,
77 (56.2%) multi. Age, Family Cohesion,
Ethnicity not reported Higher SCS, less NSSI r = −.19 (p < .001) Self‐Compassion
Victimisation associated with NSSI at t2. Interaction
Self‐compassion weakened relationship. Bullying/SCS
Interaction SCS and peer victimisation
Β‐.61, SE B = .30, β= ‐.15, p = .041
Self‐compassion not predictive of NSSI.
Jiang, You, Zheng Adolescents Cross‐sectional SCS (Neff, NSSI. 91 (13.8%) engaged in NSSI. Females more Chi‐square
et al. (2017), China N = 658 2003a, b) likely to engage in NSSI, 17.8% vs. MANCOVA
QA = 4 ( f = 264, 40.1%) 11.86%; chi sq (1, N = 607) = 4.18, Mediation

(Continues)
7
 8

TABLE 1 (Continued)

Study, Analysis
Measures Key findings Covariates
country, quality
Outcome
assessment (QA)
score Sample Study design Self‐compassion Measure
Item asking presence or p = .041, NSSI in 12m younger than Univariate tests
absence NSSI 12m those with no NSSI.
Mean age: 13.58 NSSI group lower family attachment and SCS Mediation—gender,
scores (p < .001). NSSI group lower feelings age
trust, communication, and closeness than C.
NSSI (mean = 2.97significantly lower levels
of self‐compassion, F(1, 504) = 35.56,
p < .001,.07; no hist group (mean = 3.37)
Range: 11–16 Attachment and NSSI; self‐compassion
Ethnicity not reported mediated the relationship maternal/paternal
closeness and NSSI. Also mediated the
relationship between peer communication/
closeness and NSSI.
Jiang, You, Ren, et al. Adolescents N = 606 Cross‐sectional SCS (Neff, NSSI/NSSIT Group breakdown: C 422 (154 f); NSSIT 98 Chi‐Square
(2017), China (f = 38.8%) *authors 2003a, b) (39f); NSSI 86 (42F)
QA = 4 did not report n. Item asking presence or Females more likely than male NSSI (n = 42) MANCOVA group ×
absence NSSI/NSSIT 12m 17.87% vs. 11.86% (n = 44); chi sq gender post hoc
(2,N = 606) = 4.27, p = .039. Tukey
Mean age: 13.58 No gender diffs NSSIT.
Range: 11–16 C vs. NSSI‐ significant differences Age
(p < .001) all SCS subscales
Ethnicity not reported C vs. NSSIT significant differences
(p < .001) all negative SCS subscales
NSSI vs NSSIT; NSSI significant lower
common humanity (m = 3.27 vs. 3.55,
p < .01) and self‐kindness (m = 3.06 vs.
3.38, p < .001) than NSSIT.
Rabon et al. (2018), Students N = 356 Cross‐sectional SCS‐sf (Raes Combined suicidal ideation Self‐compassion correlated with wellness, Correlations, serial
USA (f = 242, 68%) et al., 2011) and suicide attempts and negative correlation with SBQ‐R mediations
QA = 2 and depressive.
Mean age: 21.44 Carried out serial mediation. Indirect None
Range: not reported SBQ‐R (Osman et al., 2001) mediation; greater self‐compassion
White = 83.1% associated with lower depression,
Black/African in turn lower SBQ‐R score.
American = 8.5%
Asian = 4.2%
Other = 2%
CLEARE

(Continues)
ET AL.
TABLE 1 (Continued)
CLEARE

Study, Analysis
Measures Key findings Covariates
ET AL.

country, quality
Outcome
assessment (QA)
score Sample Study design Self‐compassion Measure
Multiracial = 1.1%
Hispanic = 0.6%
Refused = 0.3%
Native American = 0.3%
Tanaka, Wekerle, Adolescents 117 Cross‐sectional SCS (Neff, Suicide attempts Lower SCS score greater association Correlations, chi‐square
Schmuck, and (F = 55%) 2003a, b) Item asking presence with SA (r = .3, p < .05). Significant (high vs. low
Paglia‐Boak or absence 12m associations found between childhood self‐compassion)
Mean age:18.1
(2011), Canada emotional and physical abuse (but not regression
Range: 16–20
QA = 4 sexual abuse) and lower self‐compassion.
White = 27%,
Chi‐square: greater proportion of
Black = 31.3% Age, gender
people reporting low SCS score and
Dual/multiple
SA 16.4% vs. high SCS score 4.8%
ethnicity = 27.8% 2‐ Emotional abuse
(p < .05).
Q score
3 Physical abuse
4 emotional neglect
5 SCS score
Xavier, Pinto‐ Adolescents 643 Cross‐sectional SCS (Neff, NSSI Males higher self‐compassion and Correlations
Gouveia, and (F = 332, 51.6%) 2003a, b) lower NSSI.
Cunha (2016) RTSHIA (Vrouva, Fonagy, Self‐compassion significantly correlated T‐Tests
Portugal Mean age: 15.24, Fearon, & Roussow, with depression (r = −.64), NSSI Path Analysis Testing
QA = 5 range: 12–18 2010; Portuguese (r = −.33), and daily hassles (r = −.34). Moderation Effect
version: Xavier, Cunha, Self‐Comp.
SCS subscales:
Pinto‐Gouveia, &
Ethnicity: not reported Self‐kindness accounted 23% variance Moderation: Gender
Paiva, 2013)
NSSI; interaction term depression and
self‐kindness significant, but
self‐kindness and daily hassles not
significant. Mindfulness 24% variance
NSSI; interaction term depression
and mindfulness significant, but not
significant mindfulness and daily hassles
All negative subscales significant and
24%/25% accounted for SCS had
moderating effect on depression and NSSI;
SCS buffers against depression and NSSI

Abbreviations: ANOVA, analysis of variance; BSSI, Beck scale for suicidal ideation; C, no history of any suicidality; MANCOVA, multivariate analysis of covariance; NLE, negative life events; NSSI, nonsuicidal self‐
injury; NSSIT, nonsuicidal self‐injurious thoughts; RTSHIA, Risk‐taking and Self‐harm Inventory for Adolescents; SA, history of suicide attempt; SB, suicidal behaviours (not specified/multiple constructs mea-
sured); SBQ‐R, Suicidal Behaviours Questionnaire‐R; SC, self‐compassion; SCS; Self‐Compassion Scale; SCS‐sf; Self‐Compassion Scale Short Form; SF, self‐forgiveness; SH, any self‐harm regardless of intent;
SI, history of suicide ideation; SNAP‐2, Schedule for Nonadaptive and Adaptive Personality‐2; VA, values affirmation.
9
10 CLEARE ET AL.

3.6 | Self‐compassion, self‐harm, and self‐harm Self‐compassion was associated with better peer and familial
ideation relationships (Jiang, You, Zheng, et al., 2017) including greater
feelings of maternal (B = .20, SE = 0.05, p < .001) and paternal
Individuals with no history of self‐harm (Gregory et al., 2017; Hayes closeness (B = .18, SE = 0.04, p < .001). Greater closeness was in turn
et al., 2016) reported higher self‐compassion. Additionally, self‐harm associated with lower NSSI (maternal, OR = −1.22, SE = 0.29, p < .001;
groups scored lower on the positive subscales and higher on the neg- paternal, OR = 1.21, SE = 0.29, p < .001). The relationship between
ative subscales of the SCS than control groups. Chang et al. (2017) peer communication (B = .14, SE = 0.07, p = .032), peer closeness
reported small associations between the subscales (r = −.2 to r = −.26, (B = .21, SE = 0.04, p < .001), and NSSI (OR = −1.48, SE = 0.29,
positive subscales; r = .26 to r = .28, negative subscales) and suicidal p < .001) was fully mediated by self‐compassion.
behaviours (effect sizes: positive, r2 = 5.3; negative, r2 = 7.3). The
strength of association between self‐compassion and suicidal ideation
or NSSI ranged from r2 = 3.6 to r2 = 10.9 (Jiang et al., 2016 and Xavier 3.8 | Quantitative studies of self‐forgiveness
et al., 2016, respectively). Lower self‐compassion was associated with
higher suicidal ideation (d = −0.64, p < .001; Collett et al., 2016) and sui- Seven studies investigated the relationship between self‐forgiveness
cide attempts (r = −.3, p < .05; Tanaka et al., 2011), with 16.4% of indi- and self‐harm or suicidal ideation (see Table 2 for details). All studies
viduals with low self‐compassion reporting suicide attempts compared were carried out in the United States, were cross‐sectional, and used
with 4.8% of those with higher self‐compassion. self‐report measures. A range of populations was examined: student
In the experimental study, history of self‐harm was associated with (n = 2), community (n = 2), adolescent (n = 1), military (n = 1), and older
lower score on the SCS and state self‐compassion than the controls at adults (n = 1).
baseline (Gregory et al., 2017). Following a values affirmation task, the
self‐harm group showed greatest increases in state self‐compassion
and increased pain sensitivity; they reported the discomfort sooner 3.9 | Quality assessment
and rated it as more painful than the control condition, indicating that
increasing self‐compassion may increase sensitivity to pain and, there- Methodology quality assessment scores ranged from 2 to 7 (low to
fore, may be protective in NSSI. high quality), with six of the studies scoring under 5. All the studies
were cross‐sectional, and although two studies (Bryan, Theriault, &
Bryan, 2015; Westers et al., 2012) used validated outcome measures,
3.7 | Self‐compassion and risk factors for self‐harm all studies were self‐reports. Measures of self‐forgiveness were used
and self‐harm ideation in three studies (Bryan et al., 2015; Cheavens et al., 2016; Westers
et al., 2012); the others used single or two items. None of the studies
Higher self‐compassion was repeatedly associated with lower levels of reported power calculations and subsequently scored “0” on this cate-
risk factors for suicidal ideation and self‐harm, including lower depres- gory. However, all but one study (Nsamenang, Webb, Cukrowicz, &
sive symptoms in two studies (r = −.37, p < .05; Tanaka et al., 2011; Hirsch, 2013) included a comparison group with no self‐harm or sui-
d = −0.73, p < .001; Collett et al., 2016). Similarly, in serial mediation cidal ideation. The study that had the highest quality score (7) was
analyses, Rabon et al. (2018) found self‐compassion was directly and by Bryan and colleagues (Bryan et al., 2015), which used the Self Inju-
indirectly (through depressive symptoms and wellness behaviours) rious Thoughts and Behaviours Interview (SITBI; Nock et al., 2007) to
related to suicidal behaviours. Specifically, self‐compassion was assess the presence of suicidal ideation and suicide attempts in active
related to lower depressive symptoms, which in turn, were associated and veteran military personnel currently enrolled in college.
with greater engagement in wellness behaviours, and this was sequen-
tially associated with less suicidal behaviour. Xavier et al. (2016) found
self‐compassion mediated the relationship between daily hassles and 3.10 | Sample characteristics
NSSI in adolescents. The authors also found that five of the subscales
(not common humanity) contributed to around a quarter of the vari- The collated sample size was 1,329, with a mean age of 35 years old
ance in NSSI (self‐kindness, r = 23%, B = −.09, p = .028; mindfulness,
2
(range = 12–78 years). Overall, 57% (n = 758) of participants were
r2 = 24%, B = −.08, p = .038; self‐judgement, r2 = 25%, B = .12, female; however, whereas the majority of studies were composed of
p = .009; isolation, r2 = 24%, B = .11, p = .012; over‐identification with 70–78% female participants, Bryan et al.’s study sample was 69% male
thoughts, r2 = 25%, B = .14, p = .002). (Bryan et al., 2015). Four of the samples were predominantly White
Self‐compassion partially mediated the relationship between nega- (81.4%, Bryan et al., 2015; 93%, Chang et al., 2014; 93%, Cheavens
tive life events in the last 12 months and suicidal behaviours when et al., 2016; and 94%, Nsamenang et al., 2013). Participants in the
gender was controlled for, F (7,323) = 7.18, p < .001 (Chang et al., remaining three studies were from diverse ethnic backgrounds, and
2017), and weakened the relationship between bullying and NSSI White/Caucasians made up 17% and 19% (Hirsch et al., 2011 and
(B = − .61, SE = .30, β = ‐.15, sr = .001, p = .041) at time 2 when time
2
Hirsch et al., 2012, respectively) and 56.7% of the samples (Westers
1 NSSI was controlled for (Jiang et al., 2016). et al., 2012).
TABLE 2 Self‐forgiveness quantitative studies
CLEARE

Study, country, Measures Analysis

quality assessment
ET AL.

(QA) score Sample Study Design Self‐forgiveness Outcome Measure Key Findings Covariates
Bryan et al (2015), Military services active Cross‐sectional SF‐HSF (Thompson Suicidal ideation Group breakdown: SA = 31 (7.1%), SI = 129 Correlations, Anovas,
USA and veterans enrolled et al., 2005) and attempts (29.5%), C = 278 (63.5%). Significant Regressions
QA = 8 in college 476 difference in SF scores between groups: Age, Gender, Trauma
SITBI (Nock,
(M = 69%) Lowest SF (M = 22.97, SD = 7.47) reported History, Post Trauma
Holmberg,
SA, SI significant higher SF (M = 27.90, Stress (Pts), Veteran
Mean age: 36.2 Photos, &
SD = 7.38), C highest (M = 31.23, Status, Depression
Range: 19–78 Michel, 2007)
SD = 6.40). Regressions: SF differentiated
Ethnicity:
SA from C (OR) = 0.85, [0.80, 0.90],
Caucasian = 81.4%
p = .001) and SI (OR = 0.91 [0.86, 0.96],
African
p = .001). SF also differentiated SI from C
American = 6.1%
(OR = 0.93 [0.90, 0.96], p = .001).
Native
Covariates included SF still differentiated
American = 3.2%
SA from C (AOR) = 0.90 [0.84, 0.97],
Asian = 2.5% Pacific
p = .008), but not SI from C (AOR = 0.97
Islander = 1.1%
[0.93, 1.01], p = .111). Multinomial logistic
Dual/multi = 10.8%
regressions. SF negatively correlated with
PTS, depression severity, SI (r = −.29) and
SA (r = −.26) p < .05). SF significant
predictor of PTS (adjusted age, gender,
military versus veteran status, and
depression; .131, p = .001), F(4,
407) = 37.587, p = .001, _R2 = .180.
Chang, Kahle, Yu, and Community sample 101 Cross‐sectional Two items: Combined suicidal SF significant negative association with SB. Correlations mediations
Hirsch (2014), USA (F = 71%) BMMRS (Fetzer ideation and SB significant negative association with SF.
QA = 2 Mean age: 42.18 Institute, 2003) suicide attempts SF indirect effect on Domestic abuse‐ > SB
Range: 18–64 relationship. SF partial mediation domestic
Ethnicity: abuse and SB relationship (β = .20, p < .05).
White = 93% SBQ‐R (Osman SB (β = .13, NS); forgiveness of self None
et al., 2001) (Δβ = .07) accounted from mediation.
Inclusion of SF accounted for 34%
reduction of the variance in SB.
Cheavens, Older adults 91 Cross‐sectional HFS‐S (Thompson Suicide ideation SF significant negative association with SI and Correlations, regression,
Cukrowicz, (F = 75%) et al., 2005) depression moderation
Hansen, and SF moderated relationship PB and SI. PB and
Mean age: 70.4 GSIS‐SI (Heisel &
Mitchell (2016), SI highest when SF lowest. Held when
Range: 60+ Flett, 2006)
USA controlling for demographic variables and
Ethnicity:
QA = 3 depression
Caucasian = 93% PB and SI relationship strongest when SF Demographic variables
African American = 1% lowest. Models including all demographics depression
Hispanic = 6% and SF accounted for significant SI
11

(Continues)
TABLE 2 (Continued)
12

Study, country, Measures Analysis

quality assessment
(QA) score Sample Study Design Self‐forgiveness Outcome Measure Key Findings Covariates
variance. Including interaction terms;
Interaction PB and SF accounted for further
variance. Only SF remained significant
association with SI.
Hirsch, Webb, and Student 158 (F = 123, Cross‐sectional BMMRS (Fetzer Combined suicidal All forgiveness associated with SB. SF Regressions, mediations.
Jeglic (2011) USA 78%) Institute, 2003); ideation and significant negatively association with SB
QA = 3 Single item suicide attempts and depression
Mean age: 19.58 SF mediated depression and SB relationship. Age, gender, ethnicity,
Ethnicity: Mediation: Higher SF, lower SB effect. Fully religion, spirituality,
White = 17% SBQ‐R (Osman accounted for by indirect effect of depression, forgiveness of
Hispanic = 46% et al., 2001) depression (higher SF, lower depression) others, forgiveness by
Black = 23% Mediations: SF and depression predictive God,
Asian =4% of SB Forgiveness of others related to
American Indian/ lower SB regardless of depression
Alaskan native = 2% symptoms Forgiveness of others and SF
Other = 6% both predictive of SB.
Hirsch et al. (2012) Student 372 (F = 260, Cross‐sectional BMMRS (Fetzer Combined suicidal SF significant negative association with Correlations, regressions,
USA QA = 3 70%) Institute, 2003); ideation and inward anger, SB, and depression. Inward‐ moderations, age, gender,
Mean age: 19.6 Single item suicide attempts anger significantly positively associated, ethnicity, religion,
Ethnicity: White = 19% outward‐anger significantly negatively spirituality, depression,
SBQ‐R (Osman
associated with SB SF moderator of outward anger
Hispanic = 41% et al., 2001)
association between inward and outward‐
Black = 26%
directed anger and SB, in independent
Asian =6%
models. Effect persisted in a full model
American Indian/
including both inward and outward‐anger
Alaskan native = 1%
and all forgiveness subscales.
Other = 7%
Nsamenang et al Primary care, rural Cross‐sectional BMMRS (Fetzer Combined suicidal SF significant negative association with SB, Correlations, regressions,
(2013) USA QA = 2 community. Institute, 2003); ideation and depression. Thwarted belongingness and mediations,
Uninsured 101 Single item suicide attempts perceived burdensomeness were significant
(F = 71, 71%) Negatively association with SF (r.25_.58,
Mean age: 42.18 SBQ‐R (Osman and _.55, pso.001, respectively SF indirect Age, gender, ethnicity,
Range: 18–64 et al., 2001) relationship SB. Burdensomeness mediator religion, spirituality,
Ethnicity: SF (r = .25 to .28, p = .004) and of others (r.25 depression
White = 94% to.24, p = .017), not forgiveness by God,
were significantly negatively associated
with SB. dep and negatively association
with forgiveness of self (r = −.48, p = .001),
mediations
Higher Sf > lower dep/burdensomeness/t
belongingness>lower SB
CLEARE
ET AL.

(Continues)
 CLEARE
ET AL.

TABLE 2 (Continued)

Study, country, Measures Analysis

quality assessment
(QA) score Sample Study Design Self‐forgiveness Outcome Measure Key Findings Covariates
Mediation: Significant total and direct effects
for all forgiveness dimensions on SB not
observed coV; age, gender, ethnicity,
religion, spirituality, depression
indirect effect of SF on SB was statistically
significant.
Westers, Rehfuss, Adolescents 30 (F = 21, Cross‐sectional MFS (Mauger, NSSI Higher NSSI frequency associated with lower Correlations, regressions
Olson, and Biron 70%) Perry, Freeman, SF. gender
(2012), USA Mean age: 15.77 Range: & Grove, 1992) NSSI subscale of Lower SF associated with greater likelihood of
QA = 5 12–19 SITBI (Nock NSSI to get rid of unwanted feelings (ANR;
Ethnicity: et al., 2007) and adjusted r2 = 0.35, F2,27 = 8.91, p = .001.)
Caucasian = 56.7% Functional Lower SF significantly predictive of NSSI for
Hispanic = 30% assessment of APR, ANR, SPR. Latter 2 held when sex
African NSSI controlled for. SF only significant
American = 6.7% contribution to regression. SF significant
Native American = 3.3% predictor of engaging in NSSI for APR
Multiple ethnicities: (A = 0.45, p = .021), and for NSSI for SPR
3.3% (A = 0.43, p = .027). Association more
frequent NSSI and SF (r25 = .609, p = .001),
negative relationship.

Abbreviations: APR, automatic positive reinforcement; BMMRS, Brief Multidimensional Measure of Religiousness and Spirituality; C, no history of any suicidality; GSIS‐SI, Geriatric Suicide Ideation Scale; HFS‐S,
Heartland Forgiveness Scale; MFS, Mauger Forgiveness Scale; NSSIT, nonsuicidal self‐injurious thoughts; PB, perceived burdensomeness; SA, history of suicide attempt; SB, suicidal behaviours (not specified/
multiple constructs measured); SBQ‐R, Suicidal Behaviours Questionnaire‐R; SC, self‐compassion; SF, self‐forgiveness; SF‐HSF, self‐forgiveness subscale of the Heartland Forgiveness Scale; SH, any self‐harm
regardless of intent; SI, history of suicide ideation; SITBI, Self‐Injurious Thoughts and Behaviours Interview; SPR, social positive reinforcement.
13
14 CLEARE ET AL.

3.11 | Assessment of self‐forgiveness 3.14 | Self‐forgiveness and risk factors for self‐harm
and self‐harm ideation
Five measures of self‐forgiveness were used in studies, ranging from a
single‐ (Hirsch et al., 2011; Hirsch et al., 2012) or two‐item (Chang Self‐forgiveness moderated the relationship between perceived
et al., 2014) version of the Brief Multi‐Dimensional Measure of Reli- burdensomeness and suicidal ideation (Cheavens et al., 2016). Specif-
giousness and Spirituality (Fetzer Institute, 2003) and the self‐ ically, feeling a burden to others was associated with higher levels of
forgiveness subscale of the Heartland Forgiveness Scale (Thompson ideation in the presence of low self‐forgiveness even when depressive
et al., 2005) to the 15‐item self‐forgiveness subscale of the Mauger symptomology was controlled for. Hirsch et al. (2011) found that self‐
Forgiveness scale (Mauger et al., 1992). forgiveness's association with suicidal behaviours was fully mediated
by depressive symptoms. In their later study, Hirsch et al. (2012) found
that self‐forgiveness significantly moderated the relationship (t = −2.08,
3.12 | Assessment of self‐harm and self‐harm p < .05) between internal anger and suicidal behaviours (r = .35,
ideation p < .001). Chang et al. (2014) found that higher self‐forgiveness reduced
the association between domestic abuse and suicidal behaviours by
Suicidal thoughts and suicide attempts were addressed in six of the 34%, reducing the relationship to nonsignificant levels.
studies; however, four studies used the total score of the SBQ‐R
(Osman et al., 2001), so it is unclear what construct was assessed.
3.15 | Qualitative study of self‐compassion
Two studies (Bryan et al., 2015; Westers et al., 2012) employed the
SITBI (Nock et al., 2007); however, Westers et al. (2012) focussed
One qualitative study met inclusion criteria. Sutherland, Dawczyk, De
on the NSSI subscale. The final study (Cheavens et al., 2016) assessed
Leon, Cripps, and Lewis (2014) used a selective sampling methodology
suicidal ideation (Geriatric Suicide Ideation Scale; Heisel & Flett, to extract writings expressing positive components of the SCS (self‐
2006).
kindness, common humanity, and mindfulness; Neff, 2003) from
web/blog posts describing NSSI experiences (Table 3). The authors
explored the data using Interpretive Phenomenological Analysis tech-
3.13 | Self‐forgiveness, self‐harm, and self‐harm niques. A total of 170 posts were included from 27 websites (24 dis-
ideation cussions and 3 blog sites) primarily based in the United States and
the United Kingdom. Due to the nature of the study, no demographic
Associations between higher self‐forgiveness and lower NSSI, suicidal data were available, and it was not possible to determine respondent
behaviours, and suicidal ideation were found by all studies. However, residence, gender, and NSSI information (e.g., NSSI method and fre-
the strength of the relationship varied between studies. Cheavens quency) and whether the posts were written by different individuals
et al. (2016) reported a moderate relationship between higher self‐ or multiple posts were written by the same person. Multiple themes
forgiveness and lower levels of suicidal ideation (r = −.41, p < .01) in
were extracted from posts highlighting the interconnectedness of the
older adults. Moderate to weak associations were found between components. The authors reported that expressions of self‐compassion
higher self‐forgiveness and suicidal ideation and behaviours in com- were more apparent in writings associated with recovery, reflecting
munity (Nsamenang et al., 2013; r = −.28, p < .01; Chang et al., individuals' greater understanding of their NSSI experience and lower
2014; r = −.4, p < .001) and student (Hirsch et al., 2011; r = −.26, levels of distress. However, many posts were excluded from the study
p < .05; Hirsch et al., 2012; r = −.27, p < .001) samples. Similarly, as they discussed self‐criticism, which was not the focus of the
Bryan et al., (2015) found lower levels of suicidal ideation and research. Although the authors did not state the number of posts
attempts (r = −.29, r = −.26, respectively) were associated with higher excluded from the analysis, they did state that “many of the sites
self‐forgiveness. Self‐forgiveness also differentiated between control, included more than 100 entries.”
suicidal ideation, and attempt groups in regression analyses. Self‐
forgiveness still distinguished between the control and suicide attempt
group when sociodemographic characteristics (including age, gender, 4 | DISCUSSION
and current military status, i.e., veteran or active), depressive symp-
toms, trauma history, and stress were controlled for. Westers et al. Self‐compassion and self‐forgiveness are important factors to con-
(2012) examined self‐forgiveness and reasons for engaging in NSSI sider when assessing suicide risk, and this review aimed to under-
in adolescents. Lower self‐forgiveness predicted engaging in NSSI to stand this relationship further by critically evaluating the extant
get rid of unwanted feelings; to feel something rather than numb; research literature. We employed a broad search strategy in an attempt
and to communicate distress to others. The latter two functions held to be inclusive and searched for terms potentially synonymous with
when gender was controlled for. A strong negative association was self‐compassion. Our search strategy resulted in 18 studies that met
found between self‐forgiveness and NSSI frequency (r = −.61, inclusion criteria; however, there was considerable heterogeneity in
p = .01), indicating that individuals who engage in NSSI repeatedly study designs, populations, and measurement tools, rendering direct
experience lower levels of self‐forgiveness. comparison of studies difficult and precluded use of meta‐analytic
CLEARE ET AL. 15

TABLE 3 Qualitative study of compassion

Measures
Study, country, quality Analysis
assessment (QA) score Sample Study Design Self‐compassion Outcome Key findings covariates

Sutherland et al. (2014) IPA analysis of self‐ Convenience/ Guided by positive NSSI Multiple self‐compassion Not
Web‐based QA = N/A compassion themes purposeful subscales of SCS Free themes extracted from applicable
in 170 NSSI related sampling (Neff, 2003a, b) responses within posts. Self‐
posts on blog/websites compassion mostly
found in posts regarding
recovery from NSSI.

Abbreviations: IPA, Interpretive Phenomenological Analysis; NSSI, nonsuicidal self‐injury; SCS; Self‐Compassion Scale.

techniques. Self‐compassion and self‐forgiveness were repeatedly in self‐harm; O'Connor & Kirtley, 2018; O'Connor, 2011; Joiner,
found to be significantly and negatively correlated with self‐harm, sui- 2005), self‐compassion may be potentially useful in protecting
cide attempts, or ideation, although the strength of the associations vulnerable individuals.
ranged from weak (self‐compassion; r = −.19 Jiang et al., 2016) to strong However, the sample was composed of female students, making it
(self‐forgiveness; r = −.64; Bryan et al., 2015). Our findings echo those difficult to generalize the findings, particularly as evidence suggests
from related populations that have also shown associations between that females express greater compassion towards others and lower
higher levels of self‐compassion and lower psychopathology and self‐compassion (Tanaka et al., 2011; Yarnell et al., 2015). Similar
greater psychological well‐being (MacBeth & Gumley, 2012; Zessin, methodologies in other populations and balanced by gender may pro-
Dickhäuser, & Garbade, 2015). vide further valuable insights into the mechanisms underlying self‐
There are many possible reasons for the varying strength of associ- compassion.
ations, including the measures used. Measurement of self‐forgiveness One study (Collett et al., 2016) matched participants for age and
ranged from a single‐item to a 15‐item scale, and similar variation gender across a control and clinical group. However, different methods
was seen in the measurement of self‐harm, suicide attempts, and were used for data collection between the groups. Although a self‐
ideation. The majority of the self‐compassion studies used the total report, the clinical group completed measures during an appointment
SCS (Neff, 2003a, b) score. However, one of the advantages of the with their clinician, whereas the control data were collected via an
SCS is that it can also be used to give scores for the individual online participant pool. It wasn't clear whether the controls were
components of self‐compassion (Cleare et al., 2018; Neff et al., 2017). assessed for suicidality and if data collection was carried out at the
Muris and Petrocchi (2017) suggest that as the scale includes negative same time.
components that have stronger associations with psychopathology The SBQ‐R (Osman et al., 2001) was used in six studies. This scale
(r = .47 to .50) than the positive components (r = −.27 to −.34), using consists of four items assessing (a) ideation in the last 12 months, (b)
the total score may lead to an overestimation of the strength of the expressions of suicidality to another person, (c) likelihood of a future
relationship. Consequently, the authors emphasize the need for studies suicide attempt, and (d) the presence of past suicidal behaviours or
to examine the predictive value of the SCS subscales, as currently, little thoughts. Most studies reported the total score as an overall
is known about how the components interact. Concerns have been suicidality score (range 0–16), making it unclear which aspects individ-
expressed regarding the suitability of the SCS as a measure of self‐ uals were endorsing. Additionally, the inclusion of the future behav-
compassion, and investigating the components individually could help iour item potentially means that someone could score on this
clarify this. Additionally, research using prospective or experimental measure without having experienced any past suicidality.
designs that incorporate other measures of self‐compassion such as More research is required to explore how the components of
physiological measures to explore whether all the components contrib- self‐compassion and self‐forgiveness interact with established risk
ute equally to a person's self‐compassion or if one area is potentially factors for suicide and self‐harm. Several studies investigated
more important than others and when. mechanisms potentially linking self‐compassion or self‐forgiveness
Experimental studies manipulating self‐compassion under different and suicidal ideation or self‐harm (Chang et al., 2014; Cheavens
conditions are needed to improve understanding of how and et al., 2016; Hirsch et al., 2012; Nsamenang et al., 2013;
when components of self‐compassion are activated and how this can Rabon et al., 2018). Although no study found evidence of a direct
be used in clinical practice. Our review included one experimental relationship between self‐compassion or self‐forgiveness and self‐harm
study (Gregory et al., 2017) that found that the self‐compassion or suicidal ideation, all found support for indirect relationships. That
manipulation had a greater effect in the self‐harm group and increased is, higher self‐compassion or self‐forgiveness was associated with
pain sensitivity; participants reported pain faster and felt more intense lower levels of risk factors (e.g., depressive symptoms, perceived
pain than those in the control condition. As decreased sensitivity to burdensomeness, and internally directed anger); these in turn were
physical pain has been shown to be associated with increased associated with lower suicidal ideation, attempts, or self‐harm. This
likelihood that an individual who has thoughts of self‐harm or buffering effect could be a result of the development of self‐soothing
suicide self‐harm will act on their thoughts of self‐harm (i.e., engage associated with compassion (Gilbert, 2005; Gilbert, 2009).
16 CLEARE ET AL.

Sutherland et al.'s (2014) findings that expressions of self‐ self‐compassion has a role across multiple points of the IMV model,
compassion were primarily related to recovery from NSSI resonates or it may have an overarching effect on moderators throughout the
with Westers et al.’s (2012) findings that higher self‐forgiveness was pathway. Ultimately, further research is needed to establish this. In
reported by individuals who engaged in NSSI less frequently. However, brief, the literature highlights the potential usefulness of self‐
as Sutherland et al. selected posts regarding positive components of compassion and self‐forgiveness in protecting against self‐harm idea-
self‐compassion, only 170 posts were included in the analysis despite tion and self‐harm.
the authors reporting these were extracted from 27 websites, which
often contained in excess of 100 posts. The authors provided no infor-
mation about the proportion of posts included from each website or the 4.1 | Limitations and future directions
proportion of posts that discussed the negative SCS components. Neff
(2016) describes self‐compassion as requiring an interaction between Although we incorporated a range of terms synonymous with self‐
the positive and negative components of compassion and focusing compassion in our literature search, this involved a degree of subjec-
solely on the positive components may not reflect the true nature of tivity; therefore, there is a risk we omitted terms that others would
self‐compassion. have included. Conversely, whereas we included self‐forgiveness as
The majority of studies in the review were cross‐sectional, which a search term, other research groups may not have done so. It could
limits the conclusions that can be drawn regarding the direction of also be argued that we should have searched the grey literature, but
relationships between variables. As Bryan and colleagues (2015) we did not in an attempt to enhance the quality of studies included
highlighted, low self‐forgiveness could result from an individual's view in the review.
that their suicide attempt was an unforgivable act. Additionally, the included studies varied in outcome measurements
Additionally, although self‐forgiveness was associated with lower used, and there may be considerable heterogeneity within self‐harm
levels of self‐harm, it is unclear whether the measures used in the populations, and there may be considerable statistical noise in the data
studies are measures of true self‐forgiveness or whether they are herein. Future studies may wish to consider possible subgroup analy-
influenced by pseudo self‐forgiveness. Pseudo self‐forgiveness is an ses when deigning studies. For instance, there could be important dif-
unhelpful process during which individuals appear to make peace with ferences in the profiles of individuals who have engaged in self‐harm
themselves, but rather than accepting responsibility, they engage in once compared with multiple times and in individuals within these
defensive processes to avoid negative emotions such as shifting groups who express intent to die or report no intent. Future studies
blame, justifying their actions, and minimizing the impact of the event may wish to investigate differences in these subgroups.
(Enright, 1996; Fisher & Exline, 2006; Hall & Fincham, 2005; Tangney, Self‐compassion has been extensively researched in relation to
Boone, & Dearing, 2005). This is believed to result in a state of self‐ depression, anxiety, and stress. As yet, however, we have little under-
forgiveness without requiring offenders to take ownership of wrongs. standing of how the components of the SCS interact and contribute to
Similarly, caution should also be used when interpreting cross‐ a person's compassion or if one area is potentially more important
sectional mediation analyses seeking to explain causal mechanisms than another. To fully understand the relationship between self‐
(Maxwell & Cole, 2007). Despite the limited research, studies consis- compassion, risk factors, and self‐harm, future research may wish to
tently reported associations between higher levels of self‐compassion use theoretical models such as the IMV model of suicidal behaviour
or self‐forgiveness and lower levels of self‐harm or suicidal ideation. (O'Connor, 2011; O'Connor & Kirtley, 2018). This would allow studies
This echoes the findings from meta‐analyses such as those of MacBeth to be designed to investigate the role of self‐compassion within spe-
and Gumley (2012) and Zessin et al. (2015), which found associations cific circumstances and may be particularly beneficial in exploring
between higher levels of self‐compassion and lower psychopathology the mechanisms that underlie the relationship with self‐harm and
and greater psychological well‐being. As none of the studies in the how these constructs may be applied to support recovery.
review were guided by overarching frameworks around self‐harm, it Additionally, research in this area needs to move away from cross‐
is not clear where self‐compassion would be situated in the IMV sectional studies, as these limit the causative conclusions that may
model (O'Connor, 2011; O'Connor & Kirtley, 2018). However, self‐ drive intervention development. Research may wish to employ more
compassion is thought to develop during early childhood (MacBeth prospective designs to explore whether self‐compassion (or any of
& Gumley, 2012), and subsequently, it may buffer the impact of neg- the components) is predictive of self‐harm ideation or self‐harm
ative life events (Chang et al., 2017; Jiang et al., 2016). Consequently, it behaviours over time and to what extent self‐compassion is stable,
may have its effect across the different phases of the IMV model. For which would allow the investigation of the stability of these con-
example, due to its association with risk factors for self‐harm, the ame- structs over time as well as how they affect the relationship between
lioration of feelings of shame (Gilbert & Procter, 2006), and increase in risk factors and self‐harm or self‐harm ideation. Integrating innovative
social connectedness (Hutcherson et al., 2008), it is possible that self‐ technological measures such as ecological momentary assessment
compassion would be placed in the motivational part of the pathway. (Stone & Shiffman, 1994) should be considered as this would allow
Additionally, Gregory et al.'s (2017) finding of self‐compassion increas- explorations of how self‐compassion changes over time and as a func-
ing sensitivity to pain may indicate that self‐compassion is active in tion of daily stressors and mood, which would provide valuable insight
the volitional phase of the IMV model. It is possible, therefore, that into the relationship with risk factors and self‐harm. Additionally, it is
CLEARE ET AL. 17

crucial that future research investigates these relationships in different Barnard, L. K., & Curry, J. F. (2011). Self‐compassion: Conceptualizations,
populations. correlates, & interventions. Review of General Psychology, 15(4),
289–303. https://doi.org/10.1037/a0025754
Ideally, studies should employ standardized measures of self‐
Beck, A., & Steer, R. (1991). Manual for Beck scale for suicide ideation.
forgiveness and self‐harm ideation or self‐harm to allow comparability
Psychological corporation. New York: Psychological Corporation.
across studies. Research is also needed into the relationships between
Bluth, K., & Eisenlohr‐Moul, T. A. (2017). Response to a mindful
the components of self‐compassion and the impact of age and gender self‐compassion intervention in teens: A within‐person association
on its relationship with suicidal ideation and self‐harm. Additionally, of mindfulness, self‐compassion, and emotional well‐being outcomes.
frameworks such as the IMV model can guide testable pathways of Journal of Adolescence, 57(June), 108–118. https://doi.org/10.1016/
factors, which may mediate the relationship between self‐compassion j.adolescence.2017.04.001

and self‐harm. For instance, investigating potential mediating roles of Braehler, C., Gumley, A., Harper, J., Wallace, S., Norrie, J., & Gilbert, P.
(2013). Exploring change processes in compassion focused therapy in
defeat, entrapment, and self‐criticism in the self‐compassion and
psychosis: Results of a feasibility randomized controlled trial. British
self‐harm relationship would extend the knowledge base. Journal of Clinical Psychology, 52(2), 199–214. https://doi.org/
Self‐compassion and self‐forgiveness are potentially important 10.1111/bjc.12009
protective factors. Although there appear to be similarities between Bryan, A. O., Theriault, J. L., & Bryan, C. J. (2015). Self‐forgiveness,
the two constructs, studies investigating the relationship between posttraumatic stress, and suicide attempts among military personnel
and veterans. Traumatology, 21(1), 40–46. https://doi.org/10.1037/
self‐compassion and self‐forgiveness may provide further insight into
trm0000017
how these factors interact. The fact that these can be targeted and
Chan, M. K. Y., Bhatti, H., Meader, N., Stockton, S., Evans, J., O'Connor, R.
cultivated through meditation provides another potential intervention
C., … Kendall, T. (2016). Predicting suicide following self‐harm:
point to protect individuals who may be at risk of self‐harm or ideation. Systematic review of risk factors and risk scales. The British Journal
However, it is important to note that self‐compassion is not a panacea. of Psychiatry: The Journal of Mental Science, 209(4), 277–283.
For some individuals, especially those experiencing high self‐criticism, https://doi.org/10.1192/bjp.bp.115.170050

the process of developing self‐compassion can be distressing initially Chang, E. C., Kahle, E. R., Yu, E. A., & Hirsch, J. K. (2014). Understanding
the relationship between domestic abuse and suicide behavior in adults
(Gilbert & Irons, 2005) and requires a supportive, therapeutic envi-
receiving primary care: Does forgiveness matter? Social Work, 59(4),
ronment. Additionally, research needs to reflect the complexity of 315–320. https://doi.org/10.1093/sw/swu028
self‐compassion. Research into self‐compassion, including its compo- Chang, E. C., Yu, T., Najarian, A. S. M., Wright, K. M., Chen, W., Chang, O.
nents, should account for the fact that it likely has both state and trait D., … Hirsch, J. K. (2017). Understanding the association between
properties. Novel study designs should be used to evaluate how and negative life events and suicidal risk in college students: Examining
self‐compassion as a potential mediator. Journal of Clinical Psychology,
under which circumstances the different aspects of self‐compassion
73(6), 745–755. https://doi.org/10.1002/jclp.22374
and impact upon one another. This will provide greater insight into
Cheavens, J. S., Cukrowicz, K. C., Hansen, R., & Mitchell, S. M. (2016).
the mechanisms that may facilitate therapeutic change as well as a bet-
Incorporating resilience factors into the interpersonal theory of suicide:
ter understanding of who is mostly likely affected by self‐compassion. The role of hope and self‐forgiveness in an older adult sample. Journal of
The literature highlights the potential usefulness of self‐compassion Clinical Psychology, 72(1), 58–69. https://doi.org/10.1002/jclp.22230
and self‐forgiveness in relation to suicidal ideation and self‐harm; Clark, L. A. (2003). Manual for the Schedule for Nonadaptive and Adaptive
however, more research emphasis needs to be placed on the positive Personality (SNAP–2). Minneapolis, MN: University of Minnesota Press.

components of mental health and, as such, self‐compassion and self‐ Cleare, S., Gumley, A., Cleare, C. J., & O'Connor, R. C. (2018). An investigation
of the factor structure of the Self‐Compassion Scale. Mindfulness, 9(2),
forgiveness are important areas that deserve further research
618–628. https://doi.org/10.1007/s12671‐017‐0803‐1
attention.
Collett, N., Pugh, K., Waite, F., & Freeman, D. (2016). Negative cognitions
about the self in patients with persecutory delusions: An empirical
CONF LICT OF INT E RE ST
study of self‐compassion, self‐stigma, schematic beliefs, self‐esteem,
The authors declare that they have no conflict of interest. fear of madness, and suicidal ideation. Psychiatry Research, 239,
79–84. https://doi.org/10.1016/j.psychres.2016.02.043
ORCID Colonnello, V., Petrocchi, N., & Heinrichs, M. (2017). The psychobiological
foundation of prosocial relationships: The role of oxytocin in daily
Seonaid Cleare https://orcid.org/0000-0002-2728-9356 social exchanges. In Compassion: Concepts, research and applications
Andrew Gumley https://orcid.org/0000-0002-8888-938X (pp. 105–119). New York, NY, US: Routledge/Taylor & Francis Group.
https://doi.org/10.4324/9781315564296‐6

RE FE R ENC E S Critical Appraisal Skills Programme (CASP). (2017). Critical Appraisal Skills
Programme (CASP). Retrieved from casp‐uk.net/wp‐content/uploads/
Arensman, E., Griffin, E., & Corcoran, P. (2016). Self‐harm: Extent of the 2018/01/CASP‐Qualitative‐Checklist.pdf
problem and prediction of repetition. In R. C. O’Connor & J. Pirkis
Enright, R. (1996). Counseling within the forgiveness triad: On forgiving,
(Eds.), International Handbook of Suicide Prevention. Chichester: Wiley
receiving forgiveness, and self‐forgiveness. Counseling and Values,
Blackwell. https://doi.org/10.1002/9781118903223.ch3
40(2), 107–126. https://doi.org/10.1002/j.2161‐007X.1996.tb00844.x
Arimitsu, K., & Hofmann, S. G. (2017). Effects of compassionate thinking Fisher, M., & Exline, J. (2006). Self‐forgiveness versus excusing: the roles of
on negative emotions. Cognition and Emotion, 31(1), 160–167. remorse, effort, and acceptance of responsibility. Self and Identity, 5,
https://doi.org/10.1080/02699931.2015.1078292 127–146.
18 CLEARE ET AL.

Fetzer Institute. (2003). Measurement of religiousness/spirituality for use Mental Health, Religion & Culture, 15(3), 279–300. https://doi.org/
in research, 1–103. Retrieved from http://fetzer.org/sites/default/ 10.1080/13674676.2011.571666
files/images/resources/attachment/%5Bcurrent‐date%3Atiny%5D/
Hutcherson, C. A., Seppala, E. M., & Gross, J. J. (2008). Loving‐kindness
Multidimensional_Measurement_of_Religousness_Spirituality.pdf
meditation increases social connectedness. Emotion, 8(5), 720–724.
Foster, T., Gillespie, K., McClelland, R., & Patterson, C. (1999). Risk factors https://doi.org/10.1037/a0013237
for suicide independent of DSM‐III‐R Axis I disorder. Case‐control
Jazaieri, H., McGonigal, K., Jinpa, T., Doty, J. R., Gross, J. J., & Goldin, P. R.
psychological autopsy study in Northern Ireland. The British Journal of
(2014). A randomized controlled trial of compassion cultivation training:
Psychiatry: The Journal of Mental Science, 175, 175–179.
Effects on mindfulness, affect, and emotion regulation. Motivation and
Franklin, J. C., Ribeiro, J. D., Fox, K. R., Bentley, K. H., Kleiman, E. M., Emotion, 38(1), 23–35. https://doi.org/10.1007/s11031‐013‐9368‐z
Huang, X., … Nock, M. K. (2017). Risk factors for suicidal thoughts
Jiang, Y., You, J., Hou, Y., Du, C., Lin, M. P., Zheng, X., & Ma, C. (2016).
and behaviors: A meta‐analysis of 50 years of research. Psychological
Buffering the effects of peer victimization on adolescent
Bulletin, 143(2), 187–232. https://doi.org/10.1037/bul0000084
non‐suicidal self‐injury: The role of self‐compassion and family
Fredrickson, B. L., Cohn, M. A., Coffey, K. A., Pek, J., & Finkel, S. M. (2008). cohesion. Journal of Adolescence, 53, 107–115. https://doi.org/
Open hearts build lives: Positive emotions, induced through 10.1016/j.adolescence.2016.09.005
loving‐kindness meditation, build consequential personal resources.
Jiang, Y., You, J., Ren, Y., Sun, R., Liao, S., Zhu, J., & Ma, N. (2017).
Journal of Personality and Social Psychology, 95(5), 1045–1062.
Brief report: A preliminary comparison of self‐compassion between
https://doi.org/10.1037/a0013262
adolescents with nonsuicidal self‐injury thoughts and actions.
Gilbert, P. (2005). In P. Gilbert (Ed.), Compassion: Conceptualisations, Journal of Adolescence, 59, 124–128. https://doi.org/10.1016/
research and use in psychotherapy. London: Routledge. j.adolescence.2017.05.019
Gilbert, P. (2009). The compassionate mind. London: Constable. Jiang, Y., You, J., Zheng, X., & Lin, M. P. (2017). The qualities of attachment
Gilbert, P. (2017). In P. Gilbert (Ed.), Compassion: Concepts, research and with significant others and self‐compassion protect adolescents from
applications (1st ed.). New York: Routledge. https://doi.org/10.4324/ non suicidal self‐injury. School Psychology Quarterly, 32(2), 143–155.
9781315564296 https://doi.org/10.1037/spq0000187

Gilbert, P., & Choden (2013). Mindful compassion. London: Constable‐Robinson. Johnson, J., Wood, A. M., Gooding, P., Taylor, P. J., & Tarrier, N. (2011).
Resilience to suicidality: The buffering hypothesis. Clinical Psychology
Gilbert, P., & Irons, C. (2005). Focused therapies and compassionate mind
Review, 31(4), 563–591. https://doi.org/10.1016/j.cpr.2010.12.007
training shame and self‐attacking. In P. Gilbert (Ed.), Compassion:
Conceptualisations, research and use in psychotherapy (pp. 263–325). Joiner, T. (2005). Why people die by suicide. Cambridge, MA, US: Harvard
London: Routledge. University Press.
Gilbert, P., McEwan, K., Matos, M., & Rivis, A. (2011). Fears of compassion: Kiekens, G., Hasking, P., Boyes, M., Claes, L., Mortier, P., Auerbach, R. P.,
Development of three self‐report measures. Psychology and … Bruffaerts, R. (2018). The associations between non‐suicidal
Psychotherapy: Theory, Research and Practice, 84(3), 239–255. self‐injury and first onset suicidal thoughts and behaviors. Journal of
https://doi.org/10.1348/147608310X526511 Affective Disorders, 239(May), 171–179. https://doi.org/10.1016/
Gilbert, P., & Procter, S. (2006). For people with high shame and j.jad.2018.06.033
self‐criticism: Overview and pilot study of a group therapy approach. Kirby, J. N. (2017). Compassion interventions: The programmes, the
Clinical Psychology & Psychotherapy, 13, 353–379. https://doi.org/ evidence, and implications for research and practice. Psychology and
10.1002/cpp.507 Psychotherapy, 90(3), 432–455. https://doi.org/10.1111/papt.12104
Gregory, W. E., Glazer, J. V., & Berenson, K. R. (2017). Self‐compassion, Klimecki, O. M., Leiberg, S., Ricard, M., & Singer, T. (2014). Differential
self‐injury, and pain. Cognitive Therapy and Research, 41(5), 777–786. pattern of functional brain plasticity after compassion and empathy
https://doi.org/10.1007/s10608‐017‐9846‐9 training. Social Cognitive and Affective Neuroscience, 9(6), 873–879.
Hall, J. H., & Fincham, F. D. (2005). Self–forgiveness: Thestepchild of https://doi.org/10.1093/scan/nst060
forgiveness research. Journal of Social and Clinical Psychology, 24, Krawitz, R. (2012). Behavioural treatment of severe chronic self‐loathing in
621–637. people with borderline personality disorder. Part 2: Self‐compassion and
Hall, C. W., Row, K. A., Wuensch, K. L., & Godley, K. R. (2013). The other interventions. Australasian Psychiatry, 20(6), 501–506. https://doi.
role of self‐compassion in physical and psychological well‐being. org/10.1177/1039856212459586
The Journal of Psychology, 147(4), 311–323. https://doi.org/10.1080/ Laithwaite, H., O'Hanlon, M., Collins, P., Doyle, P., Abraham, L., & Porter, S.
00223980.2012.693138 (2009). Recovery after psychosis (RAP): A compassion focused
Hayes, J. A., Lockard, A. J., Janis, R. A., & Locke, B. D. (2016). programme for individuals residing in high security settings. Behavioural
Construct validity of the Self‐Compassion Scale‐Short Form among and Cognitive Psychotherapy, 37(05), 511. https://doi.org/10.1017/
psychotherapy clients. Counselling Psychology Quarterly, 29(4), s1352465809990233
405–422. https://doi.org/10.1080/09515070.2016.1138397 Leiberg, S., Klimecki, O., & Singer, T. (2011). Short‐term compassion
Heisel, M. J., & Flett, G. L. (2006). The development and initial training increases prosocial behavior in a newly developed prosocial
validation of the geriatric suicide ideation scale. The American Journal game. PLoS One, 6(3). https://doi.org/10.1371/journal.pone.0017798
of Geriatric Psychiatry, 14(9), 742–751. https://doi.org/10.1097/01. Lutz, A., Slagter, H., Dunne, J., & Davidson, R. (2008). Attention regulation
JGP.0000218699.27899.f9 and monitoring in meditation. Trends in Cognitive Sciences, 12(4),
Hirsch, J. K., Webb, J. R., & Jeglic, E. L. (2011). Forgiveness, depression, and 163–169. https://doi.org/10.1016/j.tics.2008.01.005
suicidal behavior among a diverse sample of college students. Journal of
MacBeth, A., & Gumley, A. (2012). Exploring compassion: A meta‐analysis
Clinical Psychology, 67(9), 896–906. https://doi.org/10.1002/jclp.20812
of the association between self‐compassion and psychopathology.
Hirsch, J. K., Webb, J. R., & Jeglic, E. L. (2012). Forgiveness as a moderator Clinical Psychology Review, 32(6), 545–552. https://doi.org/10.1016/
of the association between anger expression and suicidal behaviour. j.cpr.2012.06.003
CLEARE ET AL. 19

Mauger, P. A., Perry, J. E., Freeman, T., & Grove, D. C. (1992). The O'Connor, R. C. (2011). The integrated motivational‐volitional model of
measurement of forgiveness: Preliminary research. Journal of suicidal behavior. Canada: Crisis. https://doi.org/10.1027/0227‐5910/
Psychology and Christianity, 11(2), 170–180. a000120
Maxwell, S. E., & Cole, D. A. (2007). Bias in cross‐sectional analyses of O'Connor, R. C., Cleare, S., Eschle, S., Wetherall, K., & Kirtley, O. J. (2016).
longitudinal mediation. Psychological Methods, 12(1), 23–44. The integrated motivational‐volitional model of suicidal behavior: An
https://doi.org/10.1037/1082‐989X.12.1.23 update. In R. C. O'Connor & J. Pirkis (Eds.), The international handbook
of suicide prevention: Second Edition (2nd ed.) (pp. 220–240). Chichester:
Mcgehee, J. P. (2010). The development and implementation of a
Wiley‐Blackwell. https://doi.org/10.1002/9781118903223.ch13
self‐compassion intervention for adolescents: A pilot study.
O'Connor, R. C., & Kirtley, O. J. (2018). The integrated motivational‐
Muris, P. (2016). A protective factor against mental health problems in
volitional model of suicidal behaviour. Philosophical Transactions of
youths? A critical note on the assessment of self‐compassion. Journal
the Royal Society B, 373, 20170268. https://doi.org/10.1098/
of Child and Family Studies, 25(5), 1461–1465. https://doi.org/
rstb.2017.0268
10.1007/s10826‐015‐0315‐3
O'Connor, R. C., & Nock, M. K. (2014). The psychology of suicidal
Muris, P., & Petrocchi, N. (2017). Protection or vulnerability? A behaviour. The Lancet. Psychiatry, 1(1), 73–85. https://doi.org/
meta‐analysis of the relations between the positive and negative 10.1016/S2215‐0366(14)70222‐6
components of self‐compassion and psychopathology. Clinical
O'Connor, R. C., Sheehy, N. P., & O'Connor, D. B. (1999). A thematic
Psychology and Psychotherapy, 24(2), 373–383. https://doi.org/
analysis of suicide notes. Crisis, 20(3), 106–114. https://doi.org/
10.1002/cpp.2005
10.1027//0227‐5910.20.3.106
Neff, K. D. (2003a). Self‐compassion: An alternative conceptualization of
O'Connor, R. C., Smyth, R., Ferguson, E., Ryan, C., & Williams, J. M. G. (2013).
a healthy attitude toward oneself. Self and Identity, 2, 85–101.
Psychological processes and repeat suicidal behavior: A four‐year
https://doi.org/10.1080/15298860390129863
prospective study. Journal of Consulting and Clinical Psychology, 81(6),
Neff, K. (2003b). The development and validation of a scale to 1137–1143. https://doi.org/10.1037/a0033751
measure self‐compassion. Self and Identity, 2, 223–250. https://doi. Osman, A., Bagge, C. L., Gutierrez, P. M., Konick, L. C., Kopper, B. A., &
org/10.1080/15298860390209035 Barrios, F. X. (2001). The Suicidal Behaviors Questionnaire‐Revised
Neff, K., & Lamb, L. (2009). Self compassion. In S. Lopez (Ed.), The (SBQ‐R): Validation with clinical and nonclinical samples. Assessment,
encyclopedia of positive psychology (pp. 864–867). Hoboken, NJ: 8(4), 443–454. https://doi.org/10.1177/107319110100800409
Wiley Blackwell. Rabon, J. K., Sirois, F. M., & Hirsch, J. K. (2018). Self‐compassion and
Neff, K. D. (2016). The Self‐Compassion Scale is a valid and theoretically suicidal behavior in college students: Serial indirect effects via
coherent measure of self‐compassion. Mindfulness, 7(1), 264–274. depression and wellness behaviors. Journal of American College Health,
https://doi.org/10.1007/s12671‐015‐0479‐3 66(2), 114–122. https://doi.org/10.1080/07448481.2017.1382498

Neff, K. D., Tóth‐Király, I., Yarnell, L. M., Arimitsu, K., Castilho, P., Ghorbani, Raes, F., Pommier, E., Neff, K. D., & Van Gucht, D. (2011). Construction
N., … Mantzios, M. (2019). Examining the factor structure of the and factorial validation of a short form of the Self‐Compassion Scale.
Self‐Compassion Scale in 20 diverse samples: Support for use of a total Clinical Psychology and Psychotherapy, 18(3), 250–255. https://doi.
score and six subscale scores. Psychological Assessment, 31(1), 27–45. org/10.1002/cpp.702
https://doi.org/10.1037/pas0000629 Rasmussen, S. A., Fraser, L., Gotz, M., MacHale, S., MacKie, R., Masterton,
G., … O'Connor, R. C. (2010). Elaborating the cry of pain model of
Neff, K. D., Whittaker, T. A., & Karl, A. (2017). Examining the factor
structure of the Self‐Compassion Scale in four distinct populations: Is suicidality: Testing a psychological model in a sample of first‐time and
repeat self‐harm patients. British Journal of Clinical Psychology, 49(1),
the use of a total scale score justified? Journal of Personality Assessment,
15–30. https://doi.org/10.1348/014466509X415735
99(6), 596–607. https://doi.org/10.1080/00223891.2016.1269334
Ribeiro, J. D., Franklin, J. C., Fox, K. R., Bentley, K. H., Kleiman, E. M., Chang,
NICE (2004). Self‐harm. In The short‐term physical and psychological
B. P., & Nock, M. K. (2016). Self‐injurious thoughts and behaviours as
management and secondary prevention of selfharm in primary and secondary
risk factors for future suicide ideation, attemps and death. Psychol
care. Leicester (UK). doi.org/www.nice.org.uk/guidance/cg16
Med, 46(2), 225–236. https://doi.org/10.1017/S0033291715001804
NICE. (2011). Self‐harm: Longer term management. Www.Nice.Org.Uk/
Smeets, E., Neff, K., Alberts, H., & Peters, M. (2014). Meeting suffering
Guidance/Cg133. Retrieved from www.nice.org.uk/guidance/cg133
with kindness: Effects of a brief self‐compassion intervention for
Nock, M. K., Holmberg, E. B., Photos, V. I., & Michel, B. D. (2007). female college students. Journal of Clinical Psychology, 70(9),
Self‐injurious thoughts and behaviors interview: Development, 794–807. https://doi.org/10.1002/jclp.22076
reliability, and validity in an adolescent sample. Psychological Assessment, Sommers‐Spijkerman, M., Trompetter, H., Schreurs, K., & Bohlmeijer, E.
19(3), 309–317. https://doi.org/10.1037/1040‐3590.19.3.309 (2018). Pathways to improving mental health in compassion‐focused
Nsamenang, S. A., Webb, J. R., Cukrowicz, K. C., & Hirsch, J. K. (2013). therapy: Self‐reassurance, self‐criticism and affect as mediators of
Depressive symptoms and interpersonal needs as mediators of change. Frontiers in Psychology, 9(DEC), 1–12. https://doi.org/
forgiveness and suicidal behavior among rural primary care patients. 10.3389/fpsyg.2018.02442
Journal of Affective Disorders, 149(1–3), 282–290. https://doi.og/ Stone, A., & Shiffman, S. (1994). Ecological Momentary Assessment (EMA)
10.1016/j.jad.2013.01.042 in behavioral medicine. Annals of Behavioral Medicine, 16(3), 199–202.
O'Connor, D. B., Ferguson, E., Green, J. A., O'Carroll, R. E., & O'Connor, R. https://doi.org/10.1093/abm/16.3.199
C. (2016). Cortisol levels and suicidal behavior: A meta‐analysis. Sutherland, O., Dawczyk, A., De Leon, K., Cripps, J., & Lewis, S. P. (2014).
Psychoneuroendocrinology, 63, 370–379. https://doi.org/10.1016/ Self‐compassion in online accounts of nonsuicidal self‐injury: An
j.psyneuen.2015.10.011 interpretive phenomenological analysis. Counselling Psychology Quarterly,
O'connor, R. C. (2003). Suicidal behavior as a cry of pain: Test of a 27(4), 409–433. https://doi.org/10.1080/09515070.2014.948809
psychological model. Archives of Suicide Research, 7(4), 297–308. Tanaka, M., Wekerle, C., Schmuck, M. L., & Paglia‐Boak, A. (2011). The
https://doi.org/10.1080/713848941 linkages among childhood maltreatment, adolescent mental health,
20 CLEARE ET AL.

and self‐compassion in child welfare adolescents. Child Abuse and Neglect, Des Sciences Du Comportement, 40(1), 1–10. https://doi.org/10.1037/
35(10), 887–898. https://doi.org/10.1016/j.chiabu.2011.07.003 0008‐400x.40.1.1.1
Tangney, J. P., Boone, A. L., & Dearing, R. (2005). Forgiving the self: World Health Organization (2014). Preventing suicide: A global imperative.
Conceptual issues and empirical findings. In E. L. Worthington Jr. World Health Organization, 143(7). ISBN: 978 92 4 156477 9
(Ed.), Handbook of Forgiveness (pp. 143–158). New York: Routledge. Xavier, A., Cunha, M., Pinto‐Gouveia, J., & Paiva, C. (2013).
Thompson, L. Y., Snyder, C. R., Hoffman, L., Michael, S. T., Rasmussen, H. Exploratory study of the Portuguese version of the risk‐taking
N., Billings, L. S., … Roberts, D. E. (2005). Dispositionol forgiveness of and self‐harm inventory for adolescents. Atención Primaria,
self, others, and situations. Journal of Personality, 73(2), 313–359. 45(Especial Congreso), 165.
https://doi.org/10.1111/j.1467‐6494.2005.00311.x Xavier, A., Pinto‐Gouveia, J., & Cunha, M. (2016). The protective role of
Tóth‐Király, I., Bőthe, B., & Gábor, O. (2017). Exploratory structural self‐compassion on risk factors for non‐suicidal self‐injury in
equation modeling analysis of the Self‐Compassion Scale. Mindfulness, adolescence. School Mental Health, 8(4), 476–485. https://doi.org/
8(4), 881–892. https://doi.org/10.1007/s12671‐016‐0662‐1 10.1007/s12310‐016‐9197‐9
Vrouva, I., Fonagy, P., Fearon, P. R., & Roussow, T. (2010). The risk‐ taking and Yarnell, L. M., Stafford, R. E., Neff, K. D., Reilly, E. D., Knox, M. C., &
self‐harm inventory for adolescents: Development and psychometric Mullarkey, M. (2015). Meta‐analysis of gender differences in
evaluation. Psychological Assessment, 22, 852. https://doi.org/10.1037/ self‐compassion. Self and Identity, 14(5), 499–520. https://doi.org/
a0020583 10.1080/15298868.2015.1029966
Webb, J. R., Bumgarner, D. J., Conway‐Williams, E., Dangel, T., & Hall, B. B. Zessin, U., Dickhäuser, O., & Garbade, S. (2015). The relationship
(2017). A consensus definition of self‐forgiveness: Implications for between self‐compassion and well‐being: A meta‐analysis. Applied
assessment and treatment. Spirituality in Clinical Practice, 4(3), Psychology: Health and Well‐Being, 7(3), 340–364. https://doi.org/
216–227. https://doi.org/10.1037/scp0000138 10.1111/aphw.12051
Westers, N. J., Rehfuss, M., Olson, L., & Biron, D. (2012). The role of
forgiveness in adolescents who engage in nonsuicidal self‐injury.
How to cite this article: Cleare S, Gumley A, O'Connor RC.
Journal of Nervous and Mental Disease, 200(6), 535–541. https://doi.org/
Self‐compassion, self‐forgiveness, suicidal ideation, and
10.1097/NMD.0b013e318257c837
self‐harm: A systematic review. Clin Psychol Psychother.
Wohl, M. J. A., DeShea, L., & Wahkinney, R. L. (2008). Looking within:
Measuring state self‐forgiveness and its relationship to psychological 2019;1–20. https://doi.org/10.1002/cpp.2372
well‐being. Canadian Journal of Behavioural Science/Revue Canadienne