Evidence reviews for supporting people to be safe after self-harm

Included studies

Five comparative observational studies were included for this review. One of these was a retrospective cohort study (Ford 2020) and 4 were before-and-after studies (Bowers 2006, Kapur 2016, Noelck 2019 and Reen 2020). One of these studies was an ecological analysis of mental health service level changes (Kapur 2016).

The included studies are summarised in Table 2.

Three studies were conducted in the UK (Bowers 2006, Kapur 2016 and Reen 2020) and 2 in the US (Ford 2020 and Noelck 2019).

One study included children and adolescents admitted to an inpatient setting following a suicide attempt (Noleck 2019). The other studies did not specify previous self-harm attempts of the study participants: 2 of these studies included inpatients on psychiatric wards (Bowers 2006 and Reen 2020), 1 included incarcerated adult males diagnosed with a mental health condition (Ford 2020) and 1 included all individuals who died by suicide within 12 months of contact with a mental health service (Kapur 2016).

Four studies compared complex interventions to treatment as usual (Bowers 2006, Ford 2020, Noelck 2019 and Reen 2020). All of these studies included a staffing intervention component: 1 study appointed nurses with clinical expertise in acute inpatient care to change staffing attitudes and establish ward “rules and routine” (Bowers 2006); 1 study implemented regular twilight nursing shifts in addition to a structured programme of evening activities (Reen 2020); 1 study implemented a quality improvement intervention, including a regular staffing communication intervention in addition to a safety protocol and a full patient safety search (Noleck 2019); the retrospective cohort study conducted in prison settings evaluated specialised mental health units, involving multidisciplinary teams, staff training in communication and patient-centered care, in addition to daily activities (Ford 2020). One study compared several different safety interventions before and after implementation, including environmental changes (removal of ligature points) and staff training (Kapur 2016).

See the literature search strategy in appendix B and study selection flow chart in appendix C.

Excluded studies

Studies not included in this review are listed, and reasons for their exclusion are provided in appendix K.

Summary of the evidence

One study (Bowers 2006) compared a nursing staff intervention to treatment as usual on 2 psychiatric inpatient wards, in which nurses with clinical expertise in acute inpatient care were appointed to change staffing attitudes and establish ward “rules and routine”. The study found a significant decrease in the mean number of self-harm events per shift and no difference in the mean number of suicide attempts per shift in the 12 month follow-up period compared to the 3 month pre-intervention period (low quality).

One study (Ford 2020) compared specialised mental health units for prisoners diagnosed with a serious mental health disorder to standard of care single cell housing. The study did not find a significant decrease in the rate of self-injury at 30 or 60 days measured over the 38-month intervention period (low quality). Self-injury was reported as number of events per 100 person days, as multiple attempts could have been made by the same participant.

One study (Kapur 2016) compared the implementation of national policies at the service level (removal of collapsible curtain rails, removal of low-lying ligature points, and staff training in management of suicide risk) to treatment as usual before implementation of the policy for people who died by suicide up to 12 months after contact with mental health services. The study found significant reductions in the suicide incidence rate ratio after implementation of each the 3 interventions (very low quality). The number of patients exposed to each intervention was not reported.

One study (Noelck 2019) compared a quality improvement intervention, including a regular staffing communication intervention in addition to a safety protocol and a full patient safety search to standard care for children and adolescents who were hospitalised after a suicide attempt. The study reported a lower mean number of self-harm events per 100 patient days post-intervention compared to pre-intervention over an 18 month follow-up period (very low quality). The standard deviations of the means were not reported and not enough other data were reported to allow their calculation. The significance of the difference in means could not be determined.

One study (Reen 2020) compared regular twilight nursing shifts and a structured programme of evening activities to standard care for adolescents on an inpatient psychiatric ward. The study reported self-harm (reported as mean proportion of patients self-harming per month and the rate of self-harm per 100 bed days per month) during evening and non-evening periods over an 18-month follow-up period. The study reported a significantly lower mean proportion of patients self-harming per month in the post-intervention period, during both evening and non-evening periods compared to the pre-intervention period (low quality). The rate of self-harm per 100 bed days per month was also significantly lower in the post-intervention period, during both evening and non-evening periods relative to the pre-intervention period (low quality).

The following outcomes were not reported by any of the studies: service user satisfaction, quality of life, engagement with services and number of people leaving without assessment being completed.

See appendix F for full GRADE tables.

Included studies

A single economic search was undertaken for all topics included in the scope of this guideline but no economic studies were identified which were applicable to this review question. See the literature search strategy in appendix B and economic study selection flow chart in appendix G.

Excluded studies

Economic studies not included in the guideline economic literature review are listed, and reasons for their exclusion are provided in appendix J.

Economic

No economic studies were identified which were applicable to this review question.

The outcomes that matter most

Self-harm repetition, suicide and service user satisfaction were prioritised as critical outcomes by the committee. Self-harm repetition and suicide were prioritised as critical outcomes because they are direct measures of any differential effectiveness associated with the method of initial contact and captures both fatal and non-fatal self-harm. Service user satisfaction was chosen as a critical outcome due to the importance of delivering services which are centred around the patients’ experiences and because patient satisfaction is likely to influence whether the patient engages with the intervention.

The committee agreed that quality of life, engagement with services and number of people leaving without assessment being completed should be important outcomes. Engagement with after-care was chosen as an important outcome because the first contact after discharge may influence the likelihood of whether a person who has self-harmed will attend follow-up sessions, thereby influencing whether after-care will be effective. Quality of life was chosen as an important outcome as this is a global measure of well-being and may capture aspects of effectiveness of the interventions not captured by any of the other outcome measures. Engagement with services and number of people leaving without assessment being completed were included as they are important measures of adherence and acceptability of interventions.

The quality of the evidence

When assessed using GRADE methodology the evidence was found to range from low to very low quality. In all cases, the evidence was downgraded due to risk of bias as per ROBINS-I (due to unmeasured confounding variables and inability to ascertain intervention exposure and follow-up in the intervention group). In four studies, the evidence was downgraded due to indirectness because the proportion of the population that had previously self-harmed was unclear.

Imprecision and clinical importance of effects

When examining the evidence from each study the committee discussed the effect sizes and 95% confidence intervals for each outcome to determine whether the results were clinically meaningful. The committee noted that for the majority of comparisons, there was no important difference or no evidence of important difference in outcomes, as either effect sizes were small and confidence intervals crossed the line of no effect or confidence intervals could not be calculated based on the available data. There was evidence of a benefit in terms of self-harm repetition for removal of low lying ligature points and removal of non-collapsible curtain rails versus no removal, and for clinical staff training in management of suicide risk versus standard training on self-harm. The committee noted that the 95% confidence intervals were small indicating that the moderate effect estimates were precise, however, they were not confident of the clinical importance of the effect estimates as the data were from an ecological level observational study with a very serious risk of bias due to unmeasured confounding and classification of intervention exposure. There was evidence of a benefit in terms of mean number of patients self-harming per month and rate of self-harm for a ward environment intervention which aimed to establish rules and routine versus standard care. The committee noted that the size of the effect estimates were moderate to large and were relatively precise based on the width of the confidence intervals, however, they were not confident in the clinical importance of the effects due to concerns of risk of bias from unmeasured confounding, missing data and deviations from the intended intervention.

Benefits and harms

The recommendations were drafted on the evidence where possible, but due to concerns over the quality and paucity of evidence, they are in some parts supplemented with the committee’s own experience and expertise.

There was evidence on the benefits of a staffing intervention which established ward rules and routines in an inpatient psychiatric ward in terms of the mean number of self-harm events per shift. The committee agreed that due to the indirectness of the evidence, they could not make a strong recommendation about a specific staffing intervention, however, discussed the evidence within the wider context of continuity of care. The committee acknowledged the importance of minimising variations in care and ensuring that all staff are familiar with setting-specific layouts, policies and protocols and noted that this was particularly important in settings where consistency in staffing could not be ensured, for example where temporary bank staff were used.

The committee discussed the lack of evidence on the consistency and continuity of staffing personnel and based on their experience, they agreed that this was a fundamental aspect of supporting people to be safe after self-harm. While the committee acknowledged that continuity of care is important for all patients, they wanted to make a recommendation to highlight the benefits of minimising the number of staff that people who have self-harmed see, as this is particularly important for minimising distress in this population. Based on their experience, the committee noted that this might not be practical at all times or in all settings due to staffing constraints and staff shift patterns. The committee used the evidence presented in Evidence Report T to support these recommendations. The committee referred to guidance on ensuring continuity of care in the NICE guideline on patient experience in adult NHS services and the NICE guideline on babies, children and young people’s experience of healthcare.

The committee discussed the limited evidence on observation for people who have self-harmed and noted that in their experience, better outcomes were expected when observation was a therapeutic interaction which engaged the patient and built rapport. The committee stressed the importance of ensuring that clinical observation is considered an element of care which has important benefits for the patient’s recovery. For these reasons, the committee agreed it is important that all staff undertaking clinical observation of people who have self-harmed be trained in clinical observation, which includes engagement of the patient and rapport building. The committee agreed that observation of people who have self-harmed should not be carried out by untrained staff such as security guards based on the principle of the parity of esteem. The committee discussed experiences where observation by non-clinical staff was intimidating and caused distress for people who had self-harmed and agreed that there was an increased risk of harm if observation was carried out by untrained staff. These discussions were also used to inform the recommendations on assessment and care in general hospital settings.

The committee discussed safety considerations for people who have self-harmed when transferring between settings. While there was no evidence identified, based on their experience and expertise, the committee agreed it was important that care plans of people who have self-harmed were accessible to staff working in both primary and secondary care settings. The committee agreed that this would help to promote continuity of care across settings and minimise distress for the patient from variations in practice.

The committee discussed the limited evidence on the benefits of ensuring staff presence during periods in inpatient settings considered high-risk for episodes of self-harm. Together with their experience, the committee agreed that staff remaining visible and accessible during handovers and busy periods would have important benefits on patient safety.

There was very low quality evidence that removing low-lying ligature points and collapsible curtain rails had the benefit of reducing suicide rates in people who had been in contact with a mental health service in the previous 12 months. the committee agreed it is important to ensure a safe physical environment for all mental health patients, but a particular focus on environmental safety for people who have self-harmed is important, so that ways of self-harming are not needlessly accessible. The committee agreed that there was a risk that stigma surrounding self-harm could result in overly restrictive measures when assessing the safety of the environment, and agreed that the least restrictive measures should always be used depending on the person’s needs and vulnerabilities in order to preserve the person’s autonomy and dignity, and improve their experience of services. The restrictions taken would also vary between and within trusts and should be considered at the ward level. The committee also agreed the removal of items that could be used to self-harm should be considered, again based on the individual’s needs and vulnerabilities. They agreed that removing every potential item that could be used to self-harm would not be practical, and that some people would not need this precaution to be taken. As a result, in order to promote person-centred care, the committee agreed the person who has self-harmed should be involved in any decision-making regarding this.

The committee discussed the benefits of staff familiarising patients to the procedures and the physical environment when people who have self-harmed present to the emergency department or are admitted to inpatient wards. In their experience, ensuring the person is comfortable and knows how to access help reduces distress and repeat self-harm in what can be a highly distressing experience for the person. Despite the lack of evidence, the committee agreed that this is an important component of supporting people to be safe and should be carried out at the earliest opportunity.

While there was limited evidence, the committee highlighted the importance of all staff working in secondary care settings knowing what to do if they have immediate concerns about somebody’s safety, for example if the person has self-harmed or the professional is worried they might be about to. The committee noted that often non-specialist or temporary staff were not clear on communication procedures or, due to the sensitivity of the situation, were uncomfortable in raising concerns. The committee agreed that communication channels should be made clear and maintained to ensure all staff are capable of promptly raising concerns to ensure patient safety in secondary care settings.

Cost effectiveness and resource use

The committee noted that no relevant published economic evaluations had been identified and no additional economic analysis had been undertaken in this area. They recommended specific strategies that aimed to reduce the likely variation across the NHS in the current practice for delivering care for people who have self-harmed and ensure that current standards of care are consistently met across settings. The committee agreed that there was unlikely to be a significant resource impact from the recommendations made, as these are in line with the current practice in terms of continuity of care and staffing. Additionally, they highlighted that a substantial economic impact was unlikely as the recommendations made were marginally different from the previous NICE guidelines on self-harm.