pp 1–9

Mindless Suffering: the Relationship Between Mindfulness and Non-Suicidal Self-Injury


DOI: 10.1007/s12671-016-0657-y

Cite this article as:
Caltabiano, G. & Martin, G. Mindfulness (2016). doi:10.1007/s12671-016-0657-y


Non-suicidal self-injury is a complex behaviour, disturbingly prevalent, difficult to treat and with possible adverse outcomes in the long term. Previous research has shown individuals most commonly self-injure to cope with overwhelming negative emotions. Mindfulness has been shown to be associated with emotion regulation, and mindfulness-based interventions have shown effectiveness in a wide range of psychological disorders. This research explored whether lack of mindfulness or problems in mindfulness are involved in self-injury. A non-clinical sample of 263 participants (17–65 years) completed an online survey measuring self-injurious behaviours and mindfulness. Differences in levels of mindfulness between individuals with and without a history of self-injury were investigated. Analysis of variance indicated mindfulness (overall and in terms of specific facets “act with awareness”, “non-judge” and “non-react”) was significantly lower in individuals with a history of self-injury compared to those without. Pairwise comparisons revealed current self-injurers reported significantly lower mindfulness than past self-injurers and non-self-injurers, with medium effect sizes of d = 0.51 and d = 0.77, respectively. In logistic regression, low mindfulness significantly predicted self-injury (B = 0.04, p < .001). These findings have clinical implications, suggesting that mindfulness-based interventions may assist individuals to give up self-injurious behaviours and may be an important part of prevention strategies.


Non-suicidal self-injury NSSI Self-injurious behaviour Mindfulness Mindfulness-based interventions 


One of every five young adults in Australia has engaged in non-suicidal self-injury, and the 12-month prevalence is comparable to that of panic disorder, agoraphobia or generalised anxiety disorder (Martin et al. 2010; Swannell et al. 2014). Self-injury is potentially dangerous, with approximately 24,000 people hospitalised each year in Australia, and many thousands more attending hospital emergency rooms (Penrose-Wall et al. 2009). Although not primarily about suicide, a recent review of longitudinal and cross-sectional data revealed that self-injury is a strong predictor of suicide in the long term, regardless of age, gender and socioeconomic status (Hamza et al. 2012). Self-injury behaviours typically commence between 13 and 17 years of age (Klonsky and Muehlenkamp 2007; Martin et al. 2010; Nock et al. 2006).

Self-injury is “direct and deliberate destruction of one’s own body tissue in the absence of suicidal intent” (Nock and Favazza 2009) and “not a socially or culturally recognised behaviour within the individual’s community” (Favazza et al. 1989). An epidemiological study of 12,006 Australians revealed the most common self-injury methods as cutting (41%), scratching (40%), hitting a part of the body against a hard surface (37%) and punching, hitting or slapping one’s self (34%) (Martin et al. 2010), but self-injurers may use multiple methods (Klonsky and Muehlenkamp 2007).

Reasons for self-injury may include affect regulation, anti-dissociation, anti-suicide, interpersonal boundaries, interpersonal influence, self-punishment and sensation seeking (Klonsky 2007). Affect regulation is consistently reported as the most common function of self-injury (Briere and Gil 1998; Brown et al. 2002; Gratz 2003; Klonsky 2007); affect regulation theory suggests that self-injury is a means of self-soothing to cope with emotional distress (Gratz 2003; Klonsky 2009) or hostile interpersonal conflict (Crouch and Wright 2004).

Despite the prevalence and potential seriousness of self-injury, there are few empirically supported therapeutic interventions available to treat self-injury (Klonsky and Muehlenkamp 2007; Muehlenkamp 2006; Nock 2009). Several modified therapies have shown promise, including behaviour therapy (Lynch and Cozza 2009), problem-solving therapy (Newman 2009; Townsend et al. 2001), psychodynamic therapy (Levy et al. 2007; Ryle 2004), dialectical behaviour therapy (DBT) developed for borderline personality disorder (BPD) (patients suffering from suicidal and self-injurious behaviours to help improve their ability to regulate emotions) (Linehan 1993) and more recent expressive therapies like voice movement therapy (VMT), which employs voice expression and body movement exercises in a group setting to encourage self-expression and manage negative emotions (Martin et al. 2013).

However, modified therapies have either produced poor results or significant results have not been replicated (Nock 2010). An early Cochrane Collaboration meta-analysis concluded “all therapies remain experimental and the studies are too few and small to inspire full confidence in their results” (Binks et al. 2006), and more recent reviews are equally cautious (Hawton et al. 2016; Kliem et al. 2010; Ost 2008). Recent studies of DBT (Linehan et al. 2015; Pasieczny and Connor 2011) show improved quality and good results, and this may prove to be the therapy of choice.

Therapy should be driven by relevant theory and empirical study, as well as emerging consensus. Some recent randomised controlled studies (reviewed in Herbert and Forman 2012) suggest mindfulness may be central to therapies like DBT. Mindfulness is associated with emotional awareness and regulation (Erisman and Roemer 2010), and mindfulness-based interventions are effective in reducing symptoms for a range of psychological disorders (Baer 2003). Recent meta-analyses have demonstrated convincingly that mindfulness-based cognitive therapy (MBCT) (Segal et al. 2002) and mindfulness-based stress reduction (MBSR) are effective in a wide range of mental health problems (Keng et al. 2011; Klainin-Yobas et al. 2012).

Mindfulness can be developed through meditation practice to improve well-being and reduce psychological symptoms (Baer 2003; Kabat-Zinn 2003; Ospina et al. 2007). It is often considered to contain two key elements: (1) attention to observation, with awareness of thoughts, feelings and body sensations; and (2) a focus of being open and accepting of experiences (Bishop et al. 2004). A commonly cited definition is “paying attention in a particular way: on purpose, in the present moment, and non-judgementally” (Kabat-Zinn 1994, p. 4).

Baer et al. (2006) factor analysed five mindfulness questionnaires, which led to the development of a five-factor model of mindfulness. “Observing” comprises an awareness of internal sensations, emotions and thoughts and the external environment. “Describing” involves verbally identifying personal experiences. “Acting with awareness” is fully engaging in activities in the present moment, in contrast to acting on automatic pilot. “Non-judging of inner experience” involves not evaluating thoughts or emotions as good or bad. “Non-reactivity to inner experience” is allowing thoughts and feelings to come and go, without being overly consumed by them.

Trait mindfulness is associated with reduced negative affect and emotional reactivity to external events (Arch and Craske 2010; Creswell et al. 2007) and reduced cognitive rumination (Ramel et al. 2004). Emotional awareness is essential for successful affect regulation (Gratz and Roemer 2004), and mindfulness may increase emotional awareness leading to improved ability to regulate emotions (Erisman and Roemer 2010). Research suggests mindfulness is positively associated with self-compassion and negatively correlated with thought suppression, difficulties in emotion regulation, alexithymia, dissociation and experiential avoidance (Baer et al. 2006). MBSR is effective in treating avoidance, suppression or over-absorption of thoughts and emotions (Hayes and Feldman 2004; Kabat-Zinn 1990). Self-injurers report attempts at suppression regarding urges to self-injure (Klonsky and Glenn 2009). These findings suggest that increasing awareness of emotions through mindfulness may improve affect regulation. It follows that increasing mindfulness may be of benefit to self-injurers, through improved coping with negative thoughts and emotions.

We hypothesised self-injurers would report lower levels of mindfulness than those who have never engaged in self-injury. Further, we hypothesised that low levels of mindfulness would predict self-injury. Finally, we hypothesised that lower levels of mindfulness would be associated with the increased use of self-injury for the purpose of affect regulation. Within these hypotheses, the study explored each of the five facets of mindfulness individually, to determine which, if any, are involved in self-injury. The direction of effects proposed in the hypotheses are based on our understanding of available previous research. The cross-sectional design of the study meant we could not infer causality.



Of 307 initial participants, 44 did not complete further than the first measure (FFMQ), leaving 263 participants completing the study, 47 identifying as male, 215 as female and 1 as “neither”. Participant ages ranged from less than 17 years of age to over 65 years of age, with a mean age range of 18–24 years (SD = 1.07). The majority identified as university students (65%) and as Australian (67%), followed by 5% of Singaporean descent, 4% British, 3% American, 3% Mongolian and 3% New Zealand heritage.


Ethics approval was through the University of Queensland School of Psychology and Behavioural and Social Sciences Ethical Review Committee. Subjects were recruited through the first year psychology research participation scheme at the University of Queensland. In addition, invitations were placed on university electronic sites in various schools within UQ and self-injury Facebook sites. The survey was created using the Qualtrics program, which provided a URL link for all participants to access the study online. First year psychology students received course credit for participating, but all participation was anonymous and voluntary. Given the nature of self-injury questions, acknowledging the possibility of triggering an episode, and in accordance with expectations from our Ethics Committee, all psychology students were given the opportunity for face to face debrief on completion. Online participants were strongly advised to seek support if the questionnaire triggered adverse feelings and/or the urge to self-injure.


In addition to demographics, the questionnaire included standardised scales measuring mindfulness, self-injury and one question on mindfulness practice: “Do you currently practice yoga or meditation?” Given mindfulness-based activities, higher educational attainment and increased age are correlated with higher levels of mindfulness (Baer et al. 2008); these were controlled for in analyses.

The Five Facet Mindfulness Questionnaire (FFMQ) (Baer et al. 2006) is a 39-item self-report scale, measuring “observe”, “describe”, “act with awareness”, “non-judge” and “non-react” facets of mindfulness on five-point scales from one (never or very rarely true) to five (very often or always true). The observe facet contains eight positively worded items (e.g. “When I’m walking, I deliberately notice the sensations of my body moving”). The describe facet contains five positively worded items (e.g. “I am good at finding words to describe my feelings”) and three negatively worded items (e.g. “I have trouble thinking of the right words to express how I feel about things”). The act with awareness facet contains eight negatively worded items (e.g. “When I do things my mind wanders off and I’m easily distracted”). The non-judge facet contains eight negatively worded items (e.g. “I criticise myself for having irrational or inappropriate emotions”). The non-react facet contains seven positively worded items (e.g. “I perceive my feelings and emotions without having to react to them”). An overall mindfulness score is provided plus scores for each factor, with higher scores representing higher levels of mindfulness. The FFMQ (Baer et al. 2006) is based on five extensively used and validated mindfulness measures rather than just mindful awareness and is the most comprehensive questionnaire of mindfulness assessment involving the general population, providing greater content validity (Sauer et al. 2012).

The Inventory of Statements About Self-injury (ISAS, Klonsky and Olino 2008) is a self-report questionnaire measuring both a history of self-injury behaviours and functions. Part I seeks the number of times participants have performed any of the 12 listed self-injury behaviours, intentionally and without suicidal intent, within their lifetime (cutting, scratching, biting, banging head or other body part, burning, interfering with wound healing, carving, rubbing skin against a rough surface, pinching skin, sticking self with needles, pulling hair, swallowing dangerous substances). There is space to insert “other” behaviours not listed. There are questions regarding age of first self-injury, most recent episode, physical pain experienced, whether alone when self-injuring, duration of time between the urge to self-injure and acting and whether they wished to stop self-injuring.

If participants admit to even one self-injurious behaviour, they complete part II, a comprehensive list of 13 functions of non-suicidal self-injury (Klonsky and Muehlenkamp 2007), each containing three items beginning with “When I self-harm, …”. Possible functions include affect regulation (e.g. “I am calming myself down”), self-punishment (e.g. “I am expressing anger towards myself for being worthless or stupid”), anti-dissociation (e.g. “I am causing pain so I will stop feeling numb”), anti-suicide (e.g. “I am avoiding the impulse to attempt suicide”), interpersonal influence (e.g. “I am letting others know the extent of my emotional pain”), peer bonding (e.g. “I am fitting in with others”), sensation seeking (e.g. “I am doing something to generate excitement or exhilaration”), interpersonal boundaries (e.g. “I am creating a boundary between myself and others”), self-care (e.g. “I am creating a physical injury that is easier to care for than my emotional distress”), toughness (e.g. “I am demonstrating I am tough or strong”), marking distress (e.g. “I am creating a physical sign that I feel awful”), revenge (e.g. “I am trying to hurt someone close to me”) and autonomy (e.g. “I am establishing that I am autonomous/independent”). Statements are rated “not relevant”, “somewhat relevant” or “very relevant”. Both parts of the ISAS have high validity and reliability. Coefficient alphas for the interpersonal and intrapersonal scales were 0.88 and 0.80, respectively (Klonsky and Glenn 2009).

We included one additional question from the Deliberate Self-Harm Inventory (DSHI) (Gratz 2001), to determine severity of the self-injury behaviour: “Has this behaviour ever resulted in hospitalisation or injury severe enough to require medical treatment?” (yes/no).

Data Analyses

Statistical analysis was completed in SPSS 21. Little’s (1988) omnibus test (MCAR) before (χ2 (2) = .74, p = .692) and after (χ2 (2) = .22, p = .894) supported removal of responders completing less than 50% of the questionnaire. Further, missing data analysis revealed 8% of the mindfulness variable missing (9.8% “self-injury” group, 5.8% “no self-injury” group). Being less than 10%, this should not have biased our results (Bennett 2001) and provided adequate statistical power (Schlomer et al. 2010). Neither deletion nor substitution strategies were applied to the data.

Participants were grouped into two conditions: self-injury (n = 143) and no self-injury (n = 120). A chi-squared test revealed groups did not differ significantly in age (χ2 (6) = 7.62, p = .267), ethnicity (χ2 (29) = 39.89, p = .086), education (χ2 (5) = 4.49, p = .481) and yoga/meditation practices (χ2 (1) = .98, p = .285). However, the two groups did differ significantly on gender (χ2 (1) = 5.35, p = .015), with females more likely to be in the self-injury group than males. A Kolmogorov-Smirnov test of normality revealed that the FFMQ subscales of act with awareness in the self-injury group and non-judge and non-react in the no self-injury group violated assumptions of normality; parametric tests were therefore applied.

Pearson’s correlations were conducted to measure the relationship between control variables and mindfulness overall and then to determine whether lower levels of mindfulness were associated with the use of self-injury for the function of affect regulation.

Analysis of variance (ANOVA) was used to assess differences in mindfulness between self-injurers and non-self-injurers and then differences between current and past self-injurers, with follow-up pairwise comparisons. Similarly, one-way between-groups ANOVA assessed whether mindfulness differed in relation to severity of self-injury.

Logistic regression was used to determine which mindfulness measures predicted self-injurers compared to non-self-injurers. The model contained each predictor entered individually and was adjusted for age, education level and mindfulness practice. A second logistic regression containing all adjusted predictors entered simultaneously was used to determine which variables made a unique contribution to predicting self-injury. A subsequent hierarchical multiple regression analysis was conducted including both the non-judge and the non-react mindfulness facets, while controlling for age, education and mindfulness practice.


Prevalence of Self-Injury

More than half the sample reported self-injury at some point in their lives (n = 143, 54%), with over half of those (n = 74) self-injuring within the previous 12 months. More of the females self-injured (58%, n = 125) compared to males (38%, n = 18). Cutting was most common (46%, n = 51), followed by banging the head or body (28%, n = 31), wound interference (24%, n = 27), scratching (23%, n = 26), pulling hair (12%, n = 13), biting (11%, n = 12), pinching (11%, n = 12), burning (4%, n = 5), carving (3%, n = 3), sticking needles (2%, n = 2), rubbing skin against rough surface (1%, n = 1) and ingesting dangerous substances (1%, n = 1). Twenty-one participants (16%) reported injuries resulting in hospitalisation or requiring medical treatment. Most (76%) self-injured alone. Over half (53%) self-injured within 1 hour of the self-injury urge, 19% within 1–3 hours, 7% within 3–12 hours and 5% within 24 hours, with over a day before 16% self-injured.

Functions of Self-Injury

Participants in the self-injury group endorsed affect regulation frequently as a function of self-injury (M = 6.91, SD = 1.97), followed by self-punishment (M = 5.91, SD = 2.15), marking distress (M = 5.08, SD = 1.88), anti-dissociation (M = 4.83, SD = 2.07), toughness (M = 4.21, SD = 1.53), self-care (M = 4.08, SD = 1.43), anti-suicide (M = 4.06, SD = 1.73), interpersonal boundaries (M = 3.94, SD = 1.49), interpersonal influence (M = 3.89, SD = 1.28), revenge (M = 3.71, SD = 1.42), autonomy (M = 3.67, SD = 1.35), sensation seeking, (M = 3.53 SD = 1.01,) and peer bonding (M = 3.27, SD = .84).

Mindfulness and Self-Injury

Mindfulness showed significant positive relationships with age (r = .24, p = < .001), education (r = .14, p = .014) and mindfulness practice (yoga or meditation, r = .24, p = <.001), given higher levels of these variables were associated with higher levels of mindfulness.

One-way between-groups ANOVAs assessing differences in mindfulness between self-injurers and non-self-injurers showed that, as predicted, levels of mindfulness overall were significantly lower in self-injurers, as were three of five facets of mindfulnessacting with awareness, non-judge and non-react. The observe and describe facets showed no significant difference between self-injurers and non-self-injurers (Table 1).
Table 1

Group comparisons between self-injurers and non-self-injurers with effect sizes (eta squared) on participant mean ratings of mindfulness






(n = 143)

No self-injury

(n = 120)

Mindfulness overall

116.15 (17.88)

125.03 (16.24)




26.19 (5.21)

25.12 (5.50)



25.43 (6.02)

26.58 (5.74)


Act with awareness

22.83 (5.83)

24.87 (5.52)




22.05 (7.09)

25.86 (6.52)




19.66 (4.75)

22.68 (3.95)



*p < .01; **p < .001

To further understand the role of mindfulness in self-injury, we compared mindfulness overall between current self-injurers (n = 74), past self-injurers (those with a history of self-injury but not self-injuring within the previous 12 months) (n = 56) and non-self-injurers (n = 120). The one-way between-groups ANOVA revealed a significant effect of group (F (2, 228) = 12.21, p < .001, η2 = .10). Follow-up pairwise comparisons revealed current self-injurers (M = 112.28, SD = 18.72) reported significantly lower mindfulness than past self-injurers (M = 120.00, SD = 15.05, Cohen’s d = .51) and non-self-injurers (M = 125.00, SD = 16.48, Cohen’s d = .77, t(228) = − 4.07, p = .001).

The one-way between-groups ANOVA assessing whether mindfulness differed in relation to severity of self-injury showed that self-injurers admitted to hospital or requiring medical attention (n = 21) had lower mindfulness overall (M = 107.67, SD = 20.79), compared to those who did not require medical attention (n = 110) (M = 116.63, SD = 16.81) (F (116) = 4.01, p = .047, d = .43) (Table 2).
Table 2

Logistic regression containing control variables and mindfulness predictors entered individually, predicting likelihood of engaging in self-injury






95% CI





Mindfulness overall



























Act with awareness



























Controls were age, education and mindfulness practice. Coding for self-injury: yes = 0, no = 1

eB exponentiated B

*p < .01; **p < .001

The initial logistic regression analyses, adjusted for age, education level and mindfulness practice, showed that mindfulness overall, act with awareness, non-judge and non-react significantly predicted self-injury group membership. The risk of self-injury increased by 1.04 times for every one-unit decrease in mindfulness overall, by 1.1 times for act with awareness, by 1.1 times for non-judge and by 1.2 times for non-react.

In the second logistic regression to determine which variables made a unique contribution to predicting self-injury, a total of 241 cases were analysed. The full model was statistically significant, (χ2 (8) = 37.15, p = <.001), explaining between 16% (Cox & Snell R2) and 21% (Nagelkerke R2) of the variance, suggesting 70% of self-injurers were successfully predicted and 64% of the non-self-injury group, with overall accuracy of 67%. Only the non-react facet made a unique contribution to predicting self-injury group membership, with self-injury increasing by 1.2 times for every one-unit decrease in non-react.

Mindfulness and Affect Regulation

Pearson’s correlation was used to determine whether lower levels of mindfulness were associated with the use of self-injury for affect regulation. As predicted, there was a significant negative relationship between level of mindfulness overall and use of self-injury for affect regulation (r = −.30, p (one tailed) < .001). Lower levels of mindfulness were associated with increased use of self-injury for affect regulation. Mindfulness was found to account for 9% of the variance in engaging in self-injury for the function of affect regulation. The non-judge and non-react facets also showed significant negative correlations in the use of self-injury for affect regulation (r = −.33, p (one-tailed) < .001), explaining 11% of the variance, and (r = −.38, p (one-tailed) < .001), explaining 14% of the variance, respectively. The act with awareness facet was not significantly associated (r = −.14, p = .05) nor were the observe or describe facets (r = .04, p = .324; r = −.07, p = .214, respectively).

The hierarchical multiple regression analysis included both the non-judge and the non-react facets, while controlling for age, education and mindfulness practice. Bivariate correlations between the predictors and the criterion and descriptive statistics are shown in Table 3. Non-react was the only facet with a significant negative regression weight uniquely attributing 25% to the proportion of variance in affect regulation, β = −.30, t = −3.08, p = .003. The non-judge facet did not contribute significantly (β = −.18, t = −1.84, p = .068).
Table 3

Summary of hierarchical regression analysis for control variables, non-judge and non-react variable predicting the self-injury function of affect regulation


Model 1

Model 2





















Mindfulness practice




















F for change in R2



*p < .01; **p < .001


Despite high prevalence rates and risks associated with self-injury, there are currently no definitive or sufficiently well-researched interventions known to be effective in treating the behaviour (Nock 2010). Therapies available to date have been adaptations based on general theories of change rather than developed around core issues now known to specifically underpin self-injury. Managing overwhelming emotions has gained credibility as the most common, and possibly the most important, core issue (Klonsky 2007). We know that levels of mindfulness are associated with emotional awareness and regulation (Arch and Craske 2010; Erisman and Roemer 2010). We also know that therapies based on mindfulness have shown success in reducing other psychological disorders (Fjorback et al. 2011; Klainin-Yobas et al. 2012), and similar evidence for the use of mindfulness with troubled adolescents is emerging (Tan and Martin 2012; Tan and Martin 2014). However, to date, the research basis for a specific empirical link between mindfulness and self-injury has been lacking, despite one recent clinical study implicating mindfulness as a mediator in borderline personality and self-injury (Wupperman et al. 2013).

Our sample is not a random community sample. Prevalence of self-injury in our sample was high (54%) due to recruitment procedures actively seeking self-injuring subjects through online media and mentioning the focus of the research as self-injury. Recruitment procedures may have influenced the high ratio of females in the sample, given the high numbers of females studying psychology at our university. However, most community studies of self-injury do report a high preponderance of female subjects.

Despite these possible limitations, study participants endorsed all proposed functions of self-injury to differing degrees. Affect regulation was the most commonly reported reason for self-injury, followed by self-punishment, with sensation-seeking and peer bonding as the least commonly reported; this is consistent with the review by Klonsky (2007). The high percentage of participants in the study engaging in self-injury while alone (76%) supports research suggesting self-injury performed as a coping strategy is done in private (Klonsky 2009).

Mindfulness research suggests that increased age, higher educational attainment and mindfulness practice are associated with higher levels of mindfulness (Baer et al. 2008). Consistent with the literature, a weak but significant positive association between these control variables and mindfulness was found, and we subsequently controlled for these in further analyses. Analyses also investigated each facet of mindfulness (Baer et al. 2006) to determine whether specific mindfulness components provide greater benefits than others.

A major finding of our research, as hypothesised (hypothesis 1), is that individuals with a history of self-injury reported lower levels of mindfulness overall compared to non-self-injurers. The capacity to engage in the present moment, without concern for the past or future (acting with awareness), the ability to avoid evaluating feelings or thoughts as good or bad (non-judging) and the capacity to let go of thoughts (non-reacting), were significantly lower in those with a history of self-injury. Judging or being judgmental may lead to self-criticism, observed in those engaging in self-injury to punish the self (Lundh et al. 2007).

In contrast, awareness of thoughts and feelings (observing) and the ability to verbally identify one’s experience (describing) were not significantly different between the self-injurers and the comparison group. This confirms what we know clinically, which is that self-injurers are well aware of their thoughts and feelings, and most are capable of expressing their experiences.

If mindfulness and self-injury are associated, it could be expected that those with more severe self-injury might have even lower mindfulness. Analysis of our sample demonstrated exactly that. Mindfulness was significantly lower in those self-injuring to the point of requiring medical assistance, compared to those who self-injured but did not require medical assistance.

Of considerable importance, mindfulness in current self-injurers was significantly lower than in those who had ceased self-injury over a year prior. This could suggest that past self-injurers began with higher levels of mindfulness and were able to use it to help themselves. Conversely, it may be that past self-injurers learned, or had been helped to learn, mindfulness from friends or therapists, and this in turn helped them to cease self-injury. If it is the latter, this adds support to the need to further research mindfulness-specific therapies for people who self-injure.

We expected low mindfulness to predict self-injury (hypothesis 2). Our results support the hypothesis for mindfulness overall as well as for individual facets act with awareness, non-judge and non-react significantly predicting self-injury behaviour. Once again, the observe and describe facets did not predict self-injury. Further analysis revealed that the non-react facet uniquely predicted self-injury. This makes sense clinically, given research suggesting that the primary reason individuals engage in self-injury is to cope with heightened and negative emotions (Klonsky 2007), and mindfulness has been shown to be associated with reduced negative affect and emotional reactivity (Arch and Craske 2010). We believe our findings of low mindfulness levels (represented by increased judgments and reactions towards inner sensations and an inability to release worries and concerns) are consistent with, and integrate, the mindfulness and self-injury literatures.

Finally, we believe our results support hypothesis 3 that lower levels of mindfulness would be associated with increased use of self-injury to regulate emotions. The association was found with mindfulness overall, as well as the facets non-judge and non-react. This shows some support for the ideas that those not engaged in the present moment, who are concerned with past or future events and react, or over-react, to emotion released by events are more likely to engage in self-injury to control their emotions. Further analysis revealed that reacting negatively to thoughts and emotions (the non-react facet of mindfulness) could uniquely predict engaging in self-injury to regulate emotions. These findings are consistent with research revealing mindfulness is related to improved emotional awareness and regulation (Erisman and Roemer 2010) and lower levels of emotional reactivity and negative affect (Arch and Craske 2010; Creswell et al. 2007).

Strengths of this research are that we directly compared mindfulness in a large sample of those who engage in self-injury with non-self-injurers, that we could show that more serious self-injury was associated with even lower mindfulness and that having given up self-injury was associated with a level of mindfulness similar to those who had never self-injured. The research was hypothesis driven and based on numerous studies demonstrating that affect regulation is an important function of self-injury and that mindfulness research has shown solid results in improving other psychological problems such as depression, anxiety and PTSD, where affect is a key problem. Our results provide a strong fit with the literature and suggest novel theoretical and practical contributions to the clinical psychology literature.

The study has limitations. The design is cross-sectional, at one time point, and uses retrospective self-report measures that may not reproduce an accurate account of past situations, behaviours or emotions and may be influenced by current situations or beliefs. We acknowledge that the high proportion of females in the sample may have skewed our sample in a number of ways and possibly reduced generalisability of our findings to males who self-injure. Historically, this has been a problem in much of the self-injury literature. To confirm our findings or otherwise, future research will need to use a longitudinal design tracking self-injurers through the therapeutic process, with mindfulness measured at several time points.

There has been criticism that mindfulness scales may not provide a comprehensive assessment of all characteristics of mindfulness (Bergomi et al. 2013) or may not be an accurate and reliable measure of mindfulness (Grossman 2008, 2011; Van Dam et al. 2009), and individuals who lack an understanding of mindfulness concepts may find certain questionnaire items ambiguous. These issues need to be addressed in future studies.

An alternative explanation for our results is possible. It has been suggested that those engaging in repetitive and severe self-injury experience difficulties with impulse control (You et al. 2011). In future studies, it may be necessary to include a measure of impulse control alongside a mindfulness questionnaire.

Despite shortcomings, and as yet other studies to corroborate our work, we believe the current research findings imply that improving mindfulness may help to reduce or cease self-injury. The clinical importance of the results regarding individual mindfulness facets is that they may guide development of targeted and effective treatment strategies for self-injury.

Finding tailored ways to assist individuals to be in the present moment, release thoughts and feelings without becoming overwhelmed by them or reacting to them (non-react), as well as simply observing thoughts or feelings rather than evaluating as good or bad (non-judge) may be crucial to recovery. As an example, learning to not react to urges to self-injure may be critical. Our findings revealed that for 53% of participants who self-injured, less than 1 hour elapsed between the urge and engaging in self-injury. Engagement in self-injury decreased by more than half 1 hour after the urge and reduced to only 5% after 3–6 hours. This suggests it may be important to teach non-reaction to self-injurers, to help them allow thoughts and urges to pass rather than reacting with a conditioned response.

In addition, the current findings highlight the plight of those whose self-injury requires medical assistance as being most in need of targeted soundly based techniques to manage their emotions. How mindfulness-based interventions can be incorporated into emergency care, or hospital treatment, may provide a challenge. Our current results suggest that simple psychoeducation regarding verbal identification of thoughts and feelings (describe) or educating self-injury clients to gain further awareness of thoughts and feelings (observe) may at the very least be useless and possibly even counterproductive.

Getting patients to engage in therapeutic approaches for self-injury is often difficult, and more than half of all self-injurers do not seek treatment (Martin et al. 2010). If therapies are not carefully targeted, and not based in an understanding of the motivations underpinning self-injury, they may fail and increase both hopelessness and helplessness. Our hope is that the current research will go some way to assist developers of new therapies to ground their approaches in mindfulness techniques that can influence core reasons for the initial development of self-injury in the first place.

Compliance with Ethical Standards

Ethics approval was through the University of Queensland School of Psychology and Behavioural and Social Sciences Ethical Review Committee.

Copyright information

© Springer Science+Business Media New York 2016

Authors and Affiliations

  1. 1.Department of PsychiatryThe University of QueenslandHerstonAustralia

Personalised recommendations