Suicide is a tragic, but rare event. In 2012, the most recent year with firm data, Statistics New Zealand recorded 550 deaths by suicide, a rate of 12.3 per 100,000 population.1 The rareness and difficulties in classification of a complex and multivariate behaviour,2 and difficulties with the methodologies in psychological autopsies3 mean that much of the research is carried out with those who self harm or survive suicide attempts.4,5 In 2013, there were 7,267 intentional self-harm hospitalisations in New Zealand, a rate of 176.7 per 100,000 population.6 Although the demographics of completed suicide and DSH have some similarities (both are more common in younger than older people, in Māori compared with non-Māori ), however completed suicide is more common in males and DSH in females.1,6
The best predictors of completed suicide are self-injurious thoughts and behaviors. A meta-analysis of 172 studies reported the overall accuracy of these variables was poor. Sensitivity was estimated at 10–26%, specificity 86–87% and a pooled diagnostic Odd Ratio was 1.76 (1.45–2.15).7 Other risk factors for completed suicide are being a patient of a mental health or psychiatric service. A Danish study followed all patients admitted for DSH for eight years and estimated that 3% completed suicide over that period, but that the highest risk was in the first year.8 A recent meta-analysis of unassisted clinician risk classification found eight studies (N=22,499), and gave pooled estimates for sensitivity 0.31 (95% CI: 0.18–0.50), specificity 0.85 (0.75–0.92), positive predictive value 0.22 (0.21–0.23), and negative predictive value 0.89 (0.86–0.92). They concluded that clinician classification was too inaccurate to have utility.9 This reinforces the advice that risk assessment, particularly structured risk assessment, adds little to this, and that services should concentrate on therapeutic engagement.4,10 The Ministry of Health (MoH) suicide risk assessment guidelines recommend comprehensive assessments for patients presenting with DSH.11 The Royal Australian and New Zealand College of Psychiatrists recommend either brief contact interventions or intensive CBT for people who present to hospital with DSH.4
This issue contains an audit of risk assessment in patients presenting with DSH to a large hospital in the North Island.12 The authors used the 2003 MoH guidelines11 to develop a 16-item risk assessment checklist, and reviewed 376 electronic medical records of patients who were given a DSH diagnostic code after presenting to the emergency department in that hospital.
The data presented are descriptive only and there was no attempt by the authors to perform any further analyses. The reported findings suggest that many aspects of the recommended clinical assessments were incomplete. In particular, only 18% of these patients had their attitude to current and personal safety explored, 24% had their access to further means of self-harm documented and 27% asked the family about their ability to keep the patient safe, while 36% noted family and caregiver concerns. When considering the MoH guidelines, 8% of Māori patients were offered cultural support, 13% were given a script with limited dispensing (or ‘close control’) and 25% were given written information about medication, treatment plan and key contents.
This audit has flaws. There was no attempt to consider confounding factors. There was no attempt to follow these patients over time. There was no comparison group. Despite the paper being descriptive and uncontrolled, the authors make a series of recommendations: in particular that statements of risk alone are not, in their view, of an adequate standard and that all assessments should consider the factors that would increase the risk of suicide as well as protect from suicide.
There could be several explanations for these findings. As the authors note, patients’ electronic files may not accurately reflect what was discussed during assessment. The development of a therapeutic bond and trust between assessing clinician and patients, particularly when in great distress, may be difficult to capture in an electronic note. Because some assessments were undertaken late at night in the emergency department, it may be that cultural support and other services were not available on an urgent basis at all times. Some patients may not have agreed to family or whanau being contacted, which requires the clinician to carefully negotiate a shared method of assessment, and at times that will not be according to the guidelines but acceptable to the patient, their family or both.
We would add that although paying attention to cultural competency is important, the development of a therapeutic and trusting relationship is more so.
The important questions the study poses cannot be addressed by retrospective review of electronic files. The study highlights the potential for clinical documentation not complying with guidelines, no matter how well written and practical. However, documentation is but a precis of a therapeutic interaction. Given the current state of knowledge, ongoing contact is likely to make more difference to suicide death rates than a perfectly completed risk assessment tool. This requires adequate time to build a working relationship and develop a full formulation for the patient, which should, as the authors note, drive treatment. Realistically this is more likely to happen over a series of follow-up assessments by a community mental health team or in primary care rather than in an initial emergency department assessment.
Reliance on current suicide risk assessment tools to predict future suicide lacks evidential support. Development of a psychiatric formulation and building a therapeutic relationship may offer greater potential to recognise future suicidality, but involves greater duration and depth of contact with patients, and greater complexity than any assessment toll can provide. In the future, developing robust interventions for patients presenting with suicidal ideation and DSH might be a more appropriate focus for research than screening for suicide risk without such an intervention being readily available.