Background Self-harm among young people in the UK is possibly increasing but little is known about the reasons young people give for cessation and their link with gender or employment status.
Aims To investigate self-harm in young people, prevalence, methods used, motivations for starting and ceasing, service use, and how these are related to gender, parental social class and current labour market position.
Method Population-based survey of 1258 18- to 20-year-olds living in the Central Clydeside Conurbation, Scotland.
Results Both past and current rates of self-harm were highest among those outside the labour market. This group was most likely to want to kill themselves and did not cite specialist mental health services as helpful in ceasing self-harm. Those in full-time education more often self-harmed for a brief time, mainly to reduce anxiety.
Conclusions Current labour market position was a stronger predictor than parental social class or gender for self-harm, and was linked to level of severity, motivation for starting and ceasing, and service utilisation.
Self-harm is a relatively common and possibly increasing (Gunnell et al, 2000) problem among young people in the UK, affecting 7–14% at some point in their life (Hawton et al, 2002; Hawton & James, 2005; Skegg, 2005). The majority, however, do not present to statutory services (Hawton et al, 2002; Nada-Raja et al, 2003). Considerable research has explored self-harm behaviour in young people, including examination of socio-economic status, gender and individual factors such as sexual orientation and identity (Platt & Hawton, 2002; De Leo & Heller, 2004; Skegg et al, 2003; Rehkopf & Buka, 2006; Young et al, 2006). There is, however, only a single peer-reviewed study exploring the reasons for ceasing self-harm (Sinclair & Green, 2005), but potential risk factors such as gender or employment status were not investigated.
This population-based survey of 18- to 20-year-olds investigated three well-established predictors of self-harm – gender, parental social class and current labour market position – and examined how they relate to reasons for both self-harm behaviour and its cessation, and the use of social supports.
Participants were 1258 (49% of baseline sample) 18- to 20-year-olds from the longitudinal West of Scotland 16+ Study of health and lifestyles (Sweeting et al, 2005). They were originally recruited during their final year (1994) of primary school (age 11, n=2586, 93% of issued sample) and surveyed under exam-type conditions using questionnaires at ages 11, 13 and 15 years, and by personal interview at 18–20 years (2002–2004). Losses to follow-up were typical (e.g. more likely to originate from households of lower social class, have low educational involvement and belong to reconstituted or lone-parent households; Sweeting et al, 2001, 2005). As weights to adjust for attrition bias did not alter the results, we report unweighted data (weighted results are available from the authors). Participants were interviewed individually at 18–20 years of age by registered nurses using computer-assisted interviews, either at a central location (Glasgow University or participants' old secondary schools) or at home, depending upon availability and preference. The study received ethical approval from the Glasgow University Ethics Committee.
All measures, apart from parental social class – which was obtained at age 11 – were determined at age 18–20. Questions about self-harm were asked in a single section of the interview in the sequence given below.
Lifetime self-harm and method(s)
All participants were asked `Have you ever tried to hurt yourself or harm yourself deliberately?' and, if yes, what method(s) they had used from the following list: cutting (on the arm or wrists); cutting (elsewhere on the body); scratching or scoring; taking dangerous tablets or pills; hitting or punching self; slamming hands in door; burning (with cigarettes, lighter, etc); other way (please specify).
Reason for self-harming and age at onset
Participants who had self-harmed were asked `What are/were the reasons for doing this?' from the following list: to upset others; relieve anxiety; relieve anger; forget about something; make someone else take notice; punish myself; kill myself; not sure why; other reason (please specify). In addition, they were asked at what age they first started to harm themselves.
Current self-harm and awareness by health services and informal networks
Those who reported self-harm were asked whether this was in the past only, currently (in the past year) only, or past and current. Anyone who had self-harmed within the past year was asked `Who currently knows about this?' from a list including: psychiatrist or other mental health professional; doctor/general practitioner (GP); parents (either); spouse or partner; friend(s); brother(s)/sister(s) or other close family member; work or college; most people that know you fairly well; none of the above. Responses were subsequently collapsed into formal services (GP, psychiatrist and health professional) and informal networks (parents, family, spouse or friends).
Reasons for stopping
Those who admitted to self-harm but said that they were not currently doing so were asked the open response question: `Why did you stop?' We categorised the 65 valid responses into four broad categories that show considerable similarity to the narratives developed by Sinclair & Green (2005). The categories were: one off or temporary phase (e.g. `only happened once, no further thoughts of self-harm'); coped or felt better or found purpose (e.g. `felt I was coping better with things and that [there] were better ways to cope'); got professional help or help from family or friends (e.g. `went to see psychiatrist'); realised the harm to self or family or the `stupidity' of self-harm (e.g. `realised what life was worth and how much it hurt the family').
To assess service use, participants were asked to select from a list of services those they had used since age 11. We report use of Scottish services most relevant to self-harm, namely psychiatric, accident and emergency, Children's Panel (part of the Scottish Youth Justice System) and social work.
Parental social class
Parental social class was based on the occupation of the head of the household (father figure's current or previous occupation if not working or, in his absence, mother figure's current or previous occupation), mainly provided by parents during the first wave of the study (when the participants were aged 11 in 1994). Where missing, these data were supplemented by information provided by the children themselves, which we have found to be reliable (West et al, 2001). Occupations were categorised by reference to the 1990 Standard Occupational Classification (Office of Population Census and Surveys, 1991) into non-manual (occupational classes I–IIInm) and manual (classes IIIm–V). There were 63 instances in which social class was unclassifiable owing to either missing or poor information. With the exception of basic statistics, unclassifiable data for social class were treated as missing.
Current labour market position
Participants were asked a set of questions concerning education, training and employment to determine their main labour market position. This was classified into three broad groups: full-time education (higher or further education); training or work (either full- or part-time, or on a training course or scheme); and a non-labour market group, comprising unemployed (n=86), at home or with care responsibilities (n=37) and those sick or ill (n=12).
Chi-squared or Fisher's exact tests were used, as appropriate, for categorical data, and two-tailed t-tests to assess differences for age at onset. Logistic regression was used to test for potential confounding between parental social class and current economic position. Re-analysis of data omitting the 12 participants classified as sick or ill did not alter the results substantially. Owing to the relatively low frequencies and their more qualitative nature, we report only the raw numbers for some of the more exploratory analyses of young people's explanations for ceasing self-harm.
Self-harm, gender and labour market position
Table 1 shows the rates of current, past and lifetime self-harm according to gender, parental social class and current labour market position. Overall, we found a 7.1% lifetime prevalence of self-harm, with the majority self-harming in the past only. Only 1.6% were currently self-harming. Despite no statistically significant gender difference, there was a suggestion that young women were more likely to self-harm during their lifetime (8.4 v. 5.8%). Parental social class did not predict self-harm, but current labour market position was strongly related. There was a threefold increase in lifetime self-harm, and a six- to sevenfold increase for current self-harm, for the non-labour market group compared with those in work or full-time education. In a logistic regression, including all three socio-demographic variables (female gender OR=1.52, 95% CI 0.97–2.40; parental social class – manual OR=0.69, 95% CI 0.42–1.11; and current labour market position – full-time education v. training or work OR=1.47, 95% CI 0.86–2.52, full-time education v. non-labour market OR=3.93 95% CI 2.12–7.29), only current labour market position remained a significant predictor of lifetime self-harm.
Young women were more likely to report starting to self-harm at an earlier age (females: mean 15.0, s.d.=2.1 years; males: mean 16.4, s.d.=2.3; two-tailed t-test t=2.9, d.f.=84, P=0.004), but there was no significant difference in age at onset according to parental social class or current labour market position.
Methods of self-harm
Table 2 shows rates of self-harm by different methods within the sample as a whole. Cutting, scoring or scratching were the most common, followed by taking dangerous tablets; other (typically overtly violent) methods such as burning or punching self were relatively rare. A clear gender pattern emerged, with young women more likely to cut themselves or take dangerous tablets. The only difference according to parental social class was that those from manual backgrounds used other, more unusual, methods of self-harm (typically a combination of two methods). Partly reflecting the fact that those in the non-labour market group had the highest rates of self-harm, this group were most likely to have cut themselves, taken dangerous tables, or used other methods.
Reasons for self-harm
Table 3 shows the reasons cited by those who had self-harmed. Relief of anger was most commonly reported, followed by wanting to forget about something, relief of anxiety and desire to kill themselves. This confirms that the main motive behind young people's self-harm was to relieve negative emotions, with only a small minority saying they self-harmed in order to elicit attention and help from others.
Significant gender differences were found, with young women just under twice as likely to self-harm in order to forget something and nearly three times more likely to cite relief of anxiety. Parental social class and current labour market position were both related to reasons for self-harm, with those from manual backgrounds (35.9%) being much more likely than those from non-manual backgrounds (6.7%) to self–harm to kill themselves. Those from non-manual backgrounds were also more likely to be unclear about why they had harmed themselves. Those in full-time education were more likely to self-harm to reduce anxiety than those in training or work or the non-labour market group (although not significantly so), and those in the non-labour group were more likely to self-harm in order to kill themselves.
People aware of young person's self-harm
The small numbers of those who were currently self-harming (n=20) made it difficult to establish statistical patterns. According to the young people's reports, at least one person or agency was aware of their behaviour in most cases (17 out of 20). All mentioned somebody from their informal network, most usually a parent (11 out of 20). Formal services were less likely to be aware of the behaviour (9 out of 20), with GPs more likely (9 out of 20) than specialist mental heath professionals (4 out of 20) to know; no young person had told their college or work.
There were no gender differences with respect to who was aware of the self-harm but, despite the small numbers, we found significant differences in parental social class and current labour market position. Those from households with a manual worker as the head or outside the labour market were more likely to confide in a friend. Those outside the labour market were also more likely to tell a psychiatrist or mental health professional about their self-harm; indeed, none of those in full-time education or in work or training had confided in a psychiatrist or mental health professional (further details available from the authors).
Explanations for stopping self-harm
Table 4 shows the main explanations given by participants as to why they had stopped self-harming, according to gender, parental social class and current labour market position. The most frequent reason was that the young person had realised how damaging or futile it was to self-harm or how it hurt others (sometimes both). The second most frequent reason was that the behaviour was a single episode or represented a transient period of self-harm. This was closely followed by the development of a better coping strategy, feeling better (either due to circumstances or events), or finding a purpose in life. The least common explanation was gaining external help from professionals, close friends or an unspecified source. Only five young people specifically mentioned psychologists or psychiatrists.
Although the overall gender difference for self-harm was not significant, all who stated that they had ceased to self-harm because of professional help or help from friends were young women. There were no differences in relation to parental social class, but current labour market position was strongly associated with reason for cessation, although the small numbers in each category suggest caution in interpretation. Those in full-time education were more likely to attribute their self-harm to a temporary phase, particularly compared with those outside the labour market. Those in training or work were more likely to attribute stopping their self-harm to `coping better'. Half the young people outside the labour market attributed either upset to family, friends or dependants, or realising the futility/stupidity of harm to self as reasons.
Overall, no gender differences were found in service use, and the only difference according to social class was that those from a manual background were more likely to present to an accident and emergency department. Young people currently outside the labour market appeared to be the highest users of statutory services, with elevated use of psychiatric, accident and emergency, Children's Panel and social work services.
Previous studies have indicated that gender, social class and labour market position are all important predictors of self-harm (Platt & Hawton, 2002; Skegg, 2005). This study reports on variation in self-harm in relation to these three socio-demographic factors and investigates how they may be linked to the explanations given for engaging in and stopping self-harm, as well as considering implications for service provision. The prevalence estimates in this study are similar to those from other population-based studies (Hawton et al, 2002; Hawton & James, 2005; Skegg, 2005), strengthening the generalisability of our findings. There has only been one previous peer-reviewed study of reasons for cessation in a population who had previously self-harmed (Sinclair & Green, 2005). Our study, as well as confirming the four key narratives described in that clinic study, was able to link reasons for stopping self-harm with gender and current labour market position.
We found that the main motive behind most young people's self-harm was to relieve negative emotions. This is consistent with the few population-based studies, which have suggested that young people who self-harm may have limited coping strategies to deal with emotional difficulties or may be exposed to elevated stress levels. For example, the Child and Adolescent Self-Harm in Europe Study found that the most common immediate reason for self-harm was `to find relief from a terrible state of mind', or `wanting to die', but other, less common, reasons included `to punish self' and to bring their distress to others' attention (Rodham et al, 2004). Similar reasons are also given by clinic attendees (Nock & Prinstein, 2004).
Previous studies have identified gender as an important predictor of self-harm, with a higher prevalence among young women (Hawton et al, 2002; Hawton & James, 2005; Skegg, 2005). Although our study found an excess of females for lifetime self-harm, current rates were similar for both genders. Young women were more likely to state that they would self-harm to reduce anxiety but, counter to traditional gender differences, reducing anger via self-harm was unrelated to gender. In addition, young women were more likely to self-harm by cutting or taking tablets, whereas young men were more likely to use violent methods, as in other studies (Lewinsohn et al, 1996). We found no gender difference in relation to the number of young people reporting that they self-harmed to kill themselves. This contrasts with reported gender differences in suicide rates (Skegg, 2005) and may indicate that gender differences in completed suicides could be partially attributed to gender differences in the lethality of their chosen methods of self-harm. None of the young men said that either professional or more informal help was a primary factor in stopping self-harm, but 8 out of 40 young women said this was the main reason for cessation. One possible explanation is that current professional therapeutic interventions are more tailored towards women, who may find it easier to discuss emotional difficulties than men (van Beinum, 2003; Biddle et al, 2004).
Previous studies have shown that rates of suicide, attempted suicide and self-harm are related to socio-economic factors, although the relationship is by no means straightforward (Platt & Hawton, 2002; Rehkopf & Buka, 2006). In our study, contrary to previous research, the overall prevalence of self-harm was not strongly related to parental social class, although one specific reason (`killing myself') was more often cited by young people from manual social backgrounds. The lack of association with social class is unexpected but is consistent with evidence of equalisation in health among young people in contemporary society (West & Sweeting, 2004). When we compared the relative effect of parental social class with current labour market position, we consistently identified current labour market position as the more important factor for self-harm. Young people most at risk were those who were currently unemployed, sick or outside the labour market. This closely mirrors the results of a previous study of the impact of youth unemployment on suicidal behaviour among 18-year-olds in the same geographical area (West, & Sweeting, 1996).
Other results confirm the greater severity of self-harm among those outside the labour market, with nearly half (10 out of 23) explaining that their reason for self-harm was to kill themselves, and many reporting high service use, particularly of mental health services. However, service use may not have been related solely to self-harm, as this group are likely to have other psychological or behavioural problems that increase their use of statutory services. Few outside the labour market attributed their self-harm to a transitory phase, indicating a chronic problem, and none of the 12 who had ceased to self-harm said that specialist health services were useful in supporting them to stop.
For young people in education self-harm was more likely to be a transitory reaction to specific stress, such as examinations or academic pressures, which have previously been related to psychological distress in this cohort (West & Sweeting, 2003). For many in this group, self-harm might have been an adaptive coping mechanism to deal with temporary anxiety states and not something for which they felt they needed external help. This is compatible with research suggesting that young people who self-harm are less likely to use other coping strategies in times of stress (Evans et al, 2005). This is supported by a recent study which found that young people with anxiety disorders were least likely to use statutory services (Ford et al, 2006). Furthermore, those in our study who were currently self-harming, and who were in education, work or training, tended to be more secretive, nearly always concealing their behaviour from professionals, parents or friends, and not disclosing problems at work or college.
This study is restricted to 18- to 20-year-olds and therefore conclusions about self-harm in other age groups are not possible. Glasgow has both a relatively high level of deprivation and a high concentration of colleges and universities, and this may have boosted the power of this study to detect differences between students, those employed and those outside the labour market. Attrition may also be important, since it tends to disproportionally affect those from disadvantaged backgrounds. However, applying weights to adjust for this had negligible effects on the results, suggesting it was not a factor. The study relies on young people's reports of self-harm, but this itself may be socially patterned. Owing to the relative simplicity of this analysis there is always the possibility of omitting relevant variables. For instance, the small numbers did not allow investigation of the impact of affective and psychotic disorders on self-harm. Similarly, those outside the labour market were a heterogeneous group but their small numbers did not allow for a more detailed investigation of possible subgroup differences. However, despite these limitations, this study represents one of the largest population-based studies of self-harm in this age group.
The transient nature of self-harm behaviour found in young people in employment or education suggests a better clinical outcome for this group, despite their reluctance to access help. The results reported here suggest that the most acceptable supports for these two groups would be approaches that emphasise developing personal coping skills. In contrast, those young people who are unemployed, sick or not in full-time education are of greater concern. They are more likely to be engaging in chronic self-harm and to be actively trying to kill themselves. An important finding was that, in our study, 45% of young people who had self-harmed were known to their GP, compared with 4% in Australia (De Leo & Heller, 2004) and 13% in New Zealand (Nada-Raja et al, 2003). Therefore, Scottish GPs might be a means of targeting intervention. However, although a number of respondents said that they had accessed specialist services, they often had not found them particularly helpful. More effective interventions for this group of vulnerable young people are urgently required and, in particular, training and additional support for GPs. Targeting upstream causes (social disadvantage and chaotic personal circumstances) are likely to prove more effective than biomedical interventions alone (Platt et al, 2005).
We thank Shymala Nada-Raja who provided helpful comments on preliminary work. R.Y., P.W. and H.S. are supported by the MRC. M.v.B. is funded by NHS Lanarkshire and has an honorary attachment to the MRC unit in Glasgow.
- Received December 5, 2006.
- Revision received March 6, 2007.
- Accepted March 27, 2007.
- © 2007 Royal College of Psychiatrists