Non-suicidal self-injury (NSSI) is the deliberate and self-inflicted destruction of body tissue resulting in immediate damage (Nixon, Cloutier, & Jansson, 2008) and is associated with psychobehavioural problems, suicide attempts and completions (Conner, Langley, Tomaszewski, & Conwell, 2003; Gould, Shaffer, & Davies, 1990; Hawton & James, 2005; Hawton, Zahl, & Weatherall, 2003). Considering the scarcity of epidemiological data on NSSI, the current thesis explored NSSI within an epidemiological framework. Three studies were conducted.
Study 1 explored the wide variation in existing NSSI prevalence estimates (1.5%–54.8%) using a systematic review that followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Liberati et al., 2009) and Cochrane guidelines (Armstrong et al., 2007). Meta-analysis and meta-regression were used to estimate an international pooled prevalence of NSSI and to identify sources of between-study variation. Included were 119 studies of non-clinical samples reporting 128 NSSI prevalence estimates from 18 countries. Methodological factors contributing unique between-study heterogeneity were response format type (checklist or single item), incentive for participation, participant anonymity, and period prevalence (overall R2 = 51.6%, I2 = 87.1). The unadjusted pooled NSSI prevalence increased over time: 11.7%, 95% CI [4.2, 19.3] in 1990–1999; 14.7% [10.2, 19.2] in 2000–2005; and 19.3% [17.1, 21.6] in 2006–2012. After adjusting for the methodological factors above, the pooled NSSI prevalence did not increase over time: 13.7% [4.2, 23.2] in 1990–1999; 8.8% [1.9, 15.8] in 2000–2005; and 15.5% [8.5, 22.5] in 2006–2012. Overall, the adjusted pooled lifetime prevalence was 17.2% [8.0, 26.3] among adolescents (studies where the mean age was between 10–17 years), 13.4% [4.5, 22.3] among young adults (studies where the mean age was between 18–24 years), and 5.5% [1.7, 16.3] among adults (studies where the mean age was ≥ 25 years). This study showed that NSSI prevalence estimates are heavily influenced by methodological factors and the observed increase in NSSI over time may actually be due changes in research methodology over time.
Development of standardised methodology for data collection and measurement is essential to obtain reliable and valid NSSI data and this study may assist others in establishing greater methodological consistency in future research.
Study 2 was the Australian National Epidemiological Study of Self-Injury (ANESSI), the largest population-based NSSI study in the world to date. Its primary aims were to establish NSSI prevalence, nature and correlates. Telephone interviews were used to collect data from 12,006 respondents. Lifetime NSSI prevalence was 8.1%, 12-month prevalence was 2.6%. These figures are comparable to other mental health conditions (e.g., panic disorder and generalised anxiety disorder) in Australia. Prevalence was highest among 10–24 year olds (12.3%, 95% CI [4.2, 8.3]; 12-months) and similar across gender. Those reporting NSSI had higher levels of psychological distress (OR = 7.9, 99% CI [5.45, 11.48]), scored higher on suicidal ideation (OR = 9.03, 99% CI [6.17, 13.23]) reported a greater number of suicide attempts (OR = 11.33, 99% CI [8.11, 15.81]), and were more likely to use illicit substances (OR = 3.80, 99% CI [2.87, 5.02]). Half of those aged ≥ 18 years and 21.3% of 10 to 17-year-olds reporting NSSI had been diagnosed with a psychiatric disorder and emotion dysregulation was a key factor in NSSI. The study response rate (39%) was comparable to other NSSI research (Whitlock, Muehlenkamp, & Eckenrode, 2008) and typical survey research (Tourangeau, 2004). The high prevalence and suicide risk associated with NSSI warrant the inclusion of NSSI as a high risk group in Australia’s National Suicide Prevention Strategy. The significance of emotion dysregulation in NSSI underscores the need for interventions to focus on this deficit.
Using the ANESSI data, Study 3 explored the association between child maltreatment and NSSI (Boudewyn & Liem, 1995; Briere & Gil, 1998; Gratz, Conrad, & Roemer, 2002; Whitlock, Eckenrode, & Silverman, 2006; Wright, Friedrich, Cinq-Mars, Cyr, & McDuff, 2004; Zoroglu et al., 2003), focusing on variables hypothesised to mediate this relationship (dissociation, alexithymia, and self-blame). Results differed by gender. Physical abuse (OR = 2.75, 95% CI [1.68, 4.51]) and neglect (OR = 2.56, [1.65, 3.99]) independently increased the odds of NSSI among females; physical abuse (OR = 2.69, [1.44, 5.03]) increased odds of NSSI among males. Sexual abuse did not independently increase the odds of NSSI for either gender. For females, self-blame had the greatest effect on the child maltreatment–NSSI relationship (OR decreased by 14.6%, p < .001) and dissociation had the greatest effect for males (OR decreased by 12.9%, p = .003). The cross-sectional design of the study rules out establishing a causal relationship between child maltreatment and NSSI. A notable strength of the study was the simultaneous examination of multiple types of child maltreatment. These data suggest two therapeutic avenues in preventing NSSI among individuals with a history of child maltreatment: altering attributional style to reduce self-critical thinking, and improving emotion regulation capacity.