What Happens When You Tell Someone You Self-injure? The Effects of Disclosing NSSI to Adults and Peers

Penelope Hasking; Clare S. Rees; Graham Martin; Jessie Quigley


BMC Public Health. 2015;15(1039) 

In This Article



At baseline, 2637 Australian high school students (aged 12–18 years) completed questionnaires as part of a larger project e.g..[18] An additional 354 students participated in the study for the first time at Time 2 (one year later), and 152 students joined for the first time at Time 3 (2 year follow-up). Of these students, 45.3 % (n = 1424) completed questionnaires at all time points, and 30.5 % (n = 959) completed questionnaires in at least two waves of data collection, generally consistent with longitudinal studies examining suicidality.[19] Primary reasons for attrition included students not attending scheduled questionnaire administration, school transfers and student/parent withdrawal of consent. Mean age of participants at baseline was 13.93 years (SD = .99).

Most participants were born in Australia (89.3 %); 2.3 % identified as Aboriginal or Torres Strait Islander, representative of the indigenous population in Australia.[20] Participants from areas of higher socio-economic advantage were oversampled relative to the general Australian population.[21] Of the sample, 6.4 % reported that a doctor had told them they had an emotional or behavioural problem, most commonly mood and/or anxiety disorders; 25 % reported seeing a mental health professional, most commonly a counsellor.


Self-harm Behaviour Questionnaire – Part A (SHBQ).[22] On the SHBQ respondents indicate if they have ever engaged in NSSI, and if so, describe how they injured themselves, age of onset, recency, frequency and medical severity (from 'not at all serious' to 'life threatening'). NSSI was defined for respondents as "hurt yourself on purpose without trying to kill yourself". Respondents who indicated they engaged in NSSI with intent to die, or where method of NSSI was ambiguous (e.g. overdose), were not classified as engaging in NSSI (n = 22). Participants were also asked whether they had ever seriously thought about taking their life, and if they had ever tried to take their life. The SHBQ has excellent internal consistency (α = .95), including in adolescent samples.[23] Alphas for the present study were high (α = .88–.93).

Actual Help-seeking Questionnaire.[24] Participants were asked whether they had sought help in the past two weeks from each of 10 sources (friend, boy/girlfriend, parent, mental health worker etc.) for an emotional and/or behavioural problem, and asked to name the problem for which they sought help. The AHSQ is an accurate measure of past help-seeking behaviour, especially when used together with the General Help-Seeking Questionnaire.[24]

General Help-seeking Questionnaire.[25] The General Help-Seeking Questionnaire (GHSQ) assesses future intentions to seek help for a hypothetical emotional or behavioural problem. Participants indicated their intention to seek help from each of ten sources on a 7-point Likert scale ranging from 1 ("extremely unlikely") to 7 ("extremely likely"). The scale exhibits sound internal consistency (Cronbach's α = .70) and test-retest reliability (r = .86), modest predictive validity and strong convergent evidence for criterion validity.[25] The GHSQ had a Cronbach's alpha of .66 in the current sample.

Adolescent Coping Scale (ACS).[26] The short form of the ACS consists of 18 items assessing three primary factors: problem solving, reference to others and non-productive coping. The scale shows acceptable test-retest reliability, and predictive validity.[26] Cronbach's alphas in our sample were: non-productive = .74; problem-solving = .76; reference to others = .38.

Multidimensional Scale of Perceived Social Support (MSPSS).[27] Three subscales of this measure assess perceived support from family, friends and significant others.[27] Each statement is scored on a 7-point Likert scale (1 = very strongly disagree, 7 = very strongly agree). The MSPSS has strong internal consistency for adolescents with alphas .81–.92 for the three subscales, good construct validity, discriminant validity and test-retest reliability.[28] With the current sample, perceived family support, friend support and significant other support demonstrated Cronbach's alphas of .90, .91 and .93 respectively.

Brief Reasons for Living Questionnaire – for Adolescents (BRFL-A).[29] The BRFL-A is a 14-item measure that assesses reasons adolescents endorse for not ending their lives, rated on a 6-point scale (1 = not at all important; 6 = extremely important). Five factors assess: moral objections, fear of social disapproval, survival and coping beliefs (i.e. belief in being able to find other solutions to problems), responsibility to family and fear of suicide. Confirmatory factor analysis supports the 5-factor structure, and the measure has evidenced the ability to distinguish suicidal from non-suicidal adolescents.[29] In the current study: fear of disapproval α = .62; moral objection α = .67; survival and coping beliefs α = .74; family responsibility: α = .72; fear α = .71.


After receiving ethical approval from Monash University, the University of Queensland and the educational jurisdictions who oversee access to schools, schools in five Australian states/territories were invited to participate. Both single-sex and co-educational schools were approached, however, more all-girl schools than all-boy schools agreed to participate (all-girl schools = 11; all-boy schools = 4; co-educational schools = 25), resulting in an over-representation of girls in the sample (Time 1 = 68.0 %, Time 2 = 70.7 %, Time 3 = 71.2 %).[30] Information sheets and consent forms were distributed to all parents/guardians of students in the first three or four years of school. At baseline, 3117 students received parental consent to participate, a rate (21 %) consistent with previous Australian studies requiring active parental consent.[31]

To protect confidentiality and enable identification in the event responses raised concerns about immediate risk, a unique code was derived for each student who participated in the study. Participants completed the questionnaire on school grounds; researchers were present to clarify questions. Participants took approximately one hour to complete the questionnaire, and on completion, received an information pack with printed materials about mental health issues and resources in the community. The same procedure was used at all time points.

Data Analysis

Data was not missing completely at random (Little's MCAR; χ2 (4677) = 4966.58, p < .01), but attrition analyses suggested data was at least missing at random (MAR).[32] Multiple imputation was used to replace less than 10 % missing data within each wave. Analyses with imputed and complete case data revealed minimal discrepancies; imputed data are reported.

Relationships with continuous dependent variables were tested with a series of doubly multivariate analyses of variance, in which disclosure of NSSI (yes/no) was entered as a between-subjects factor and changes in the continuous variables over time as within-subjects factors. Four MANOVAs assessed the effects of disclosure and changes over time on four sets of dependent variables, grouped thematically: 1. Help-seeking behaviour (past help-seeking and future help-seeking intentions), 2. Coping strategies (problem solving, reference to others, non-productive coping), 3) Social support (from family, friends, significant others), 4. Reasons for living (moral objections, fear of social disapproval, survival and coping beliefs, responsibility to family, fear of suicide). A final mixed-model ANOVA assessed changes in severity of NSSI (assessed with the total score from the SHBQ). These were repeated, selecting only participants who had disclosed their NSSI, to assess whether there were differential consequences of confiding in peers or adults. Means across time are presented in Table 1. Multivariate effects for each MANOVA are presented in Table 2.