Currently, the most important predictor of an individual’s risk of future self-harm and suicide, is a history of previous self-harm (Carroll et al, 2014; Owens et al, 2002). However, most people who have self-harmed will not re-attempt and will not die by suicide (Owens et al, 2002).
Psychological interventions targeted at self-harm may reduce that ongoing risk: a recent Cochrane review found that longer-term CBT (cognitive behavioural therapy) and DBT (dialectical behavioural therapy) have the most evidence for reducing future self-harm repetition but there is little evidence for more acute interventions (Hawton et al, 2016). The ideal acute psychological intervention could be delivered before discharge from an acute hospital setting and would reduce the immediate risk of future self-harm.
In a new study (O’Connor et al, 2017), the authors trialed the use of a promising acute psychological intervention: the ‘Volitional Help Sheet’ (VHS), which has been shown to reduce poor coping mechanisms such as smoking and lack of physical activity (Armitage, 2008; Armitage et al, 2010). It is based on three separate but co-operative psychological principles:
- The “integrated motivational volitional model of self-harm behaviour” (O’Connor et al, 2002), which suggests that stressful life events appraised by an individual as being humiliating potentially lead to feelings of entrapment that may subsequently increase the risk of suicidal ideation dependent on various ‘motivational moderators’ (such as little social support and limited positive thoughts about the future) (O’Connor et al, 2012);
- The stages of change from Prochaska and DiClemente’s “Transtheoretical Model”, which is a 5 stage change process (pre-contemplation, contemplation, preparation, action, maintenance) on which the therapy motivational interviewing is based and is primarily used for addictions (Prochaska J and DiClemente C, 1983);
- “Implementation interventions” (if X happens, then what could you do?), which have been shown to help reduce poor coping strategies if delivered at a contextually-appropriate time and place.
For the VHS, participants were asked to complete 2 columns: 1 with potential situations, and the other with potential solutions excluding self-harm. The 3 primary outcomes determined prior to starting the study were: Could delivery of the VHS following a self-harm episode and whilst in hospital:
- Reduce frequency of any self-harm repetition in the following 6 months?
- Increase the time to any next self-harm episode?
- Be cost-effective?
The authors conducted a prospective, single-site, randomised controlled trial (RCT) at a hospital in Edinburgh, UK. Individuals who were 16+ and admitted to hospital for a suicide attempt were eligible if they had a history of previous self-harm and met a few other conditions (fluent in English, medically well-enough, not in another study).
Participants were randomly assigned on a 1:1 basis using web-based randomisation to receive either:
- VHS plus usual treatment (the intervention group) or
- Treatment as usual (the control group).
Randomisation was stratified by sex and self-reported past self-harm history. All research and clinical staff were blinded to the randomisation, except for the researcher conducting the VHS with patients. Follow-up data over 6 months was extracted from NHS information services in Scotland, and from individual patient records, but patients did not take part in a repeat clinical assessment.
Post-hoc (after conducting the trial, and outside of the trial protocol), the authors completed extra analyses. This is often done when the authors believe that there are significant results within the data, but the analysis as determined by the protocol is not bringing these to the fore.
Between 2012 and 2014, 1,308 people were potentially eligible for the study, and 518 participants were recruited. Equal numbers (259 and 259) were assigned to the intervention and the control.
The authors reported no unintended effects or adverse events in either group, but 3 patients (0.58%) died by suicide in the 6 months follow-up (1 in the intervention group and 2 in the control group).
Primary outcome analysis was conducted according to intention-to-treat as per the protocol. In intention-to-treat, every participant must remain in their randomised group regardless of the individual outcome. The authors produced a “listwise intention-to-treat” (excluding those lost to follow-up = 254/259 in interventional arm and 258/259 in control arm) and “imputed intention-to-treat” (all randomised patients = 259 in both arms).
However, the authors also undertook a series of post-hoc analyses; the authors wished to analyse the groups without 11 patients who did not receive the VHS despite being in the intervention arm, without 2 patients who completed the VHS but did not receive the 2-month booster VHS, and without those lost to follow-up. From this they produced a “listwise post-hoc” (excluding those who did not receive the VHS as planned, who did not receive the 2-month booster VHS, and who were lost to follow up = 246/259 in interventional arm and 258/259 in control arm) and an “imputed post-hoc” (only excluding those who did not receive the VHS as planned = 248/259 in interventional arm and 259/259 in control arm).
For the main analyses:
- The VHS intervention did not significantly affect the number of people who presented with repeat self-harm during the 6 months follow-up, in both list-wise and imputed intention-to-treat analyses.
- Cox proportional survival analyses revealed no significant differences between the intervention and control groups in terms of number of days to the next self-harm episode for both listwise (hazard ratio [HR] 0·81, 95% CI 0·55 to 1·20, p=0·30) and imputed intention-to-treat analyses (HR 0·80, 0·55 to 1·18, p=0·26).
- Mean total costs for NHS hospital services per person over the 6-month follow-up were not significantly different between the intervention and control groups (£513 in intervention group versus £561 in control group, 95% CI £353 to £257, p=0·76).
These outcomes did not show significance when limiting to only those with self-harm hospital admissions in the last 10 years, or those where VHS had been completed.
However, the authors did find some interesting and significant findings, predominately focussing on the subgroup of participants who had a previous self-harm hospital admission in the last 10 years:
- In this subgroup, the control group had significantly more repeat self-harm episodes during 6 months’ follow-up than the intervention group when analysed listwise (OR 2·54, 95% CI 1·06 to 6·11, p=0·037) and imputed (OR 2·46, 95% CI 1·03 to 5·89, p=0·043). In the listwise analysis, for those with admissions for self-harm in the past 10 years, repeat self-harm was significantly reduced by 69% when exposed to the VHS (p=0·0050).
- Mean costs over the follow-up period in this subgroup were lower in the intervention group who had completed the VHS (£428) compared to the control group (£717) (difference in means £289, 95% CI=–672 to 93, p=0·14).
- In terms of numbers needed to treat within this subgroup, one in every 12 patients treated should show benefit from the VHS.
Conversely, and very importantly, for those with no previous history of hospital admission for self-harm, there was no significant effect of treatment (difference in re-presentation –8·3%, 95% CI –3·51 to 20·11, p=0·17), and in fact one in 13 might be harmed (ie. might re-present with self-harm).
There was no evidence that the VHS brief intervention affected the number of, or time to, repeat self-harm episodes for all-comers in the study. However, there was evidence that the VHS might be of use in reducing the number of future self-harm episodes in participants who had a previous admission to hospital for self-harm in the past 10 years. Nonetheless, even within this subgroup, the VHS intervention did not have any effect on the time to a repeat hospitalisation for self-harm.
The mean total costs per person for NHS hospital services in the VHS intervention group over the 6 months were not statistically different from those in the control group. However, the VHS may have up to a 90% chance of being cost-effective if it is limited to those with a history of being admitted to hospital for self-harm.
These findings fit with a “personalised medicine approach” (i.e.) different approaches are required for different patients with the same presentation. This also is consistent with the recent Cochrane review (Hawton et al, 2016): more intensive psychological interventions appear to be most effective amongst those with a (perhaps more severe) history of self-harm.
However, there are few uncertainties that remain:
- What was the role, if any, of the booster VHS (that occurred at 2 months)?
- Is the VHS associated with harm in those without a previous self-harm-related hospital admission, and why might this be?
Strengths and limitations
Although this is an interesting and important study, there are several limitations, many of which are acknowledged by the authors:
- The study was powered to detect a “clinically meaningful” reduction in the proportion of people who would self-harm, but it was not powered to detect a reduction in self-harm repetition for the sub-groups (such as those with previous self-harm-related hospitalisation in the last 10 years) or to detect differences in costs.
- Furthermore, the subgroup analyses for those with a previous self-harm hospital admission were unplanned and driven by the resulting data. As with all post-hoc analyses, we must consider these results with a certain amount of caution. They are a useful way of examining serendipitous findings, but we have to beware that it does not turn into ‘data-dredging’.
- The authors did not assess repeat self-harm in the community, or presentation to A&E outside one health board. This may have missed repeat self-harm in some participants. Although this is understandable practically, if the VHS did not perhaps decrease repetition but decreased severity (to non-presenting community episodes) this would not have been detected.
- The study was performed at a single large general hospital in Scotland. The authors felt that there was no reason to question the generalisability of findings to other hospital-treated suicide attempt populations, although there is not necessarily any evidence provided to support this.
- The authors changed the protocol in terms of the “booster VHS” at 2 months after recruitment; it was planned that the participant’s completed form would be sent to them, but instead a blank form was sent. This was transparently noted as a change in the article, but it is not clear whether it was approved as an amendment by the ethics committee?
- It is concerning that the perhaps less severe end of the patient spectrum (those without a previous self-harm-related hospitalisation) may have been caused harm by the VHS. This requires urgent consideration and further exploration.
Implications for practice
Self-harm and preventing its repetition (which may lead to morbidity and mortality) is a very important aim.
Psychological interventions would seem to be an appealing option. Longer-term forms (such as DBT) are known to be helpful, but these are a limited resource and there is an unavoidable delay in patients receiving them. Shorter-term options that could be delivered before acute hospital discharge to reduce the immediate risk would be a ‘holy grail’. On the face of it, the VHS is not an obvious solution to this problem, but the authors feel that the post-hoc analyses indicate potential hope with a more directed approach at perhaps the more severe patients.
Importantly, this paper also shows how we are progressing as a scientific community to publish negative trial results (in terms of primary outcomes) in well-reputed journals. This is necessary to further general knowledge in interested communities and to prevent unnecessary duplication.
O’Connor RC, Ferguson E, Scott F, et al. A brief psychological intervention to reduce repetition of self-harm in patients admitted to hospital following a suicide attempt: a randomised controlled trial. The Lancet Psychiatry. 2017;4(6):451-460. doi:10.1016/S2215-0366(17)30129-3.
Armitage CJ. (2008) A volitional help sheet to encourage smoking cessation: a randomized exploratory trial. Health Psychology. 27 557–66.
Armitage CJ, Arden MA. (2010). A volitional help sheet to increase physical activity in people with low socioeconomic status: a randomised exploratory trial. Psychological Health. 25 1129–45.
Carroll R, Metcalfe C, Gunnell D. (2014). Hospital management of self-harm patients and risk of repetition: systematic review and meta-analysis. Journal of Affective Disorders. 168 476–483.
O’Connor RC, Rasmussen S, Hawton K (2012). Distinguishing adolescents who think about self-harm from those who engage in self-harm. British Journal of Psychiatry. 200 (4) 330-335; DOI: 10.1192/bjp.bp.111.097808
Psychosocial interventions for self-harm in adults. Cochrane Database of Systematic Reviews 2016, Issue 5. Art. No.: CD012189. DOI: 10.1002/14651858.CD012189., , , , , , , .
Owens D, Horrocks J, House A. (2002). Fatal and non-fatal repetition of self-harm. Systematic Review. British Journal of Psychiatry. 181 193–199.
Prochaska J, DiClemente C. (1983) Stages and processes of self-change in smoking: toward an integrative model of change. Journal of Consulting and Clinical Psychology. 5 390–395
- Hernán Piñera CC BY 2.0
- Photo by Jamie Taylor on Unsplash
- Photo by Ethan Hoover on Unsplash
- Photo by Oscar Sutton on Unsplash
- Moonlightbulb CC BY 2.0
Thank you for sharing this study. Hopefully it will be beneficial and have a positive effect on the continued prevention of self-harm on individuals
It was a very interesting read
Do we know whether the repeat episodes of self-harm were and different in terms of risk or severity? In a community setting, we sometimes find that improving education around self-harm can improve help-seeking (e.g. an intervention might mean people are more likely to re-present in a clinical setting in order to receive appropriate care – and this may be more positive response than self-harming but not help seeking – I’m not sure if it would have been possible to capture this depth of data). Additionally, methods of self-harm, how risky it was, whether the patient felt in control and / or how well they looked after injuries etc could also be impacted by intervention…. I may be clutching at straws, but it would be interesting to know more about this. At a community level I am finding that frequency of self-harm is too crude a measure and that sometimes we are seeing a reduce in severity or risk associated with the harm and / or improved help seeking which may be relevant but not relegated by our current measures which are often as simple as ‘did they re-present in specialist healthcare’ or ‘did they self-harm again? / how often?’ etc…