Most who die by suicide were considered to be at low or no risk. We need a new framework for understanding the dangers
Isabel wanted to escape the pain she has lived with for years. She wasn’t sure what would come after death, but she knew that it would be better than her life at that moment. A few hours later, after her husband found her surrounded by empty packets of medication and a bottle of vodka, I am preparing to assess her in the emergency department.
From her notes, I see that Isabel has been seeing a community mental health team for the past year because of depression. In the first appointment, they picked up on some worrying use of binge-drinking to cope with her feelings, but this hasn’t been brought up again. I also see she had an appointment last week to increase the dose of her antidepressant, and that the team updated her risk assessment. I check the risk of suicide: low.
My heart sinks. ‘Low’ tells me nothing about Isabel, and it doesn’t explain why she has tried to end her life. ‘Low’ has left me more confused than when I started.
At the moment, in many parts of the world, anyone who is assessed by a mental health professional will have a suicide risk assessment in their first appointment. If they are seen by someone new – for example, in an emergency department – the assessment will be done again. On the basis of this, a patient is typically classified as being at a low, medium or high risk of suicide, or no risk at all.
The trouble is that these categories are extremely poor at predicting if an individual will die by suicide. Research shows that 85 per cent of people who died by suicide while under the care of mental health services were deemed to have ‘low’ or ‘no’ immediate risk at their final assessment. At the same time, only 5.5 per cent of people labelled at ‘high risk’ of suicide will go on to die that way. Expert enquiries have now clearly stated that these ‘categorical’ risk-assessment tools shouldn’t be used to make clinical decisions, such as whether to admit someone to hospital.
Despite this, the majority (85 per cent) of mental health services in the United Kingdom are using this type of suicide risk assessment. This approach is also common in other countries, including the United States. This is partly because they’re quick and straightforward to use: a simple series of tick boxes (a free-text section is usually provided at the end, but it’s optional). In addition, these simple categories make it easier to collect data for research purposes. But speed and ease don’t count for much when the tool isn’t accurate, and therefore doesn’t actually help patients or communication between clinicians.
The problem is that categorical risk assessments rely too much on population-wide risk factors, rather than taking individual circumstances into account. A decision about whether a person is at high risk, and therefore what level of care is needed, is based on broad factors such as their age and whether they’re unemployed. But it’s the individual details that tell us something about how badly a person needs help, and what we can do about it.
Suicide is a highly complex human behaviour that involves myriad personal factors and a large amount of uncertainty. How can placing someone in a risk category ever be useful when people are pushed and pulled out of these categories every day by life events and the fluctuating symptoms of mental illness? Consider the chance involved in a parent leaving the medication cabinet unlocked, or the death of a loved one in a car accident, and how these events would impact on someone’s risk of suicide. Additionally, population-wide risk factors for suicide largely tell us about a person’s lifetime risk of suicide, but in clinical practice we need to make judgments about the coming days, weeks and months. A final issue is that ‘low’ and ‘high’ are comparative words: ‘low’ and ‘high’ compared with what? This is never specified in a risk-assessment tool.
However we assess risk, we mustn’t let human kindness and compassion fall by the wayside
Despite these challenges, we can’t become nihilists, believing that there’s nothing we can do to reduce the risk of suicide. This would create excuses for substandard care and ignore the vast body of research into risk factors for suicide and interventions that can help. For instance, acute alcohol intoxication is associated with a more than six-fold increased risk of suicidal acts. Understanding this would prompt clinicians to ask about recent binge-drinking, and not just about chronic alcohol use, in their assessments. This, in turn, would create more opportunities to provide support for alcohol misuse, which has been shown to decrease suicide rates.
What we need is a different framework for understanding suicide risk. One such framework is risk formulation, which involves creating a prose-style narrative about a person’s risk. A comprehensive version of this was proposed by the American academic clinicians Anthony Pisani, Daniel Murrie and Morton Silverman in 2016. Instead of simply categorising a patient as low, medium or high, risk formulation creates a management plan that’s highly individualised and fully integrated within the assessment process. Instead of asking What is the level of risk?, it asks What is contributing to the risk, and what can we do to help? Critically, the patients themselves are involved in understanding their risk and what can be done going forwards – an example of the ‘co-production’ recommended in the Royal College of Psychiatrists’ 2020 report on suicide and self-harm.
With this approach, the clinician gathers information from the patient’s file, and the patients themselves, to determine four different aspects of risk:
- Risk status. This part is still categorical (low, medium, high) but the difference is that the clinician must specify which subgroup the person is being compared against. For example, I might write that Isabel’s level of risk is higher than most people with depression due to her history of binge-drinking. This step is informed by ‘static’ risk factors such as age, gender and whether the person has a history of self-harm, and also protective factors such as whether the person has children or strong religious beliefs against suicide.
- Risk state. In this step, we consider how a person’s current risk level compares with their own risk level at a previous timepoint – such as at their last appointment, or averaged across multiple past appointments. Risk state is informed by ‘dynamic’ factors: recent stressful events such as a job loss or breakup, recent suicide attempts, the patient’s current levels of distress, and their relationship with mental health services. This tells us if there have been significant changes in this person’s individual circumstances that could put them at an increased risk of suicide.
- Available resources to draw on (or safety plan). This describes what patients could do when they’re next having a tough time or feeling actively suicidal. It’s tailored to the patient’s specific circumstances and can include practical advice on making the home safer, calming activities, distractions and sources of support.
- Foreseeable changes that could exacerbate risk. In this step, we note any potential issues such as an upcoming change of care for a young person, unstable housing, financial difficulties and relationships problems. This helps us understand whether there might be problems on the horizon, which can then feed back into the safety plan above. For example, if we know that a patient often binge-drinks when feeling stressed or disrupted, and has a history of suicidal behaviour after binge-drinking, then a plan can be created for helping to cope differently with potential upcoming stress.
Risk formulation as presented here is still a relatively new concept. However, it’s already gaining traction in a national curriculum for suicide prevention in the US. In the UK, 15 per cent of services already use formulation-based tools, but they’re often developed locally. National guidelines around suicide risk formulation are needed, as these would provide detail about a ‘gold standard’ approach to formulation, ensuring consistency and quality across services.
We also need empirical studies that compare the rates of suicide in people who had a risk formulation with those who had a categorical risk assessment. If these show that risk formulation does indeed improve patient outcomes, then a cultural shift is needed. I know, from doing them myself, that writing formulations takes a lot of time and work, but I also know that they give other clinicians a richer understanding of an individual’s needs. Systems of practice in healthcare are deeply ingrained and take years to change, not helped by IT systems already set up to use categorical risk-assessment tools. It will take time to build the massive support required to embed risk formulation into standard psychiatric practice.
However we assess risk, we mustn’t let human kindness and compassion fall by the wayside. The so-called ‘assessment’ is a therapeutic exercise as much as it is a diagnostic one. When clinicians meet a patient in the emergency department, that person has possibly never told anyone what they’re going through, and they’re sharing these feelings with a total stranger. Clinicians should be warm, empathic and nonjudgmental. They should validate the person’s distress and help them make sense of what’s happening to them. When I meet Isabel in the chaos of an emergency department at 3am, I need to remember that she hasn’t come here for me to give her a diagnosis, or tell her how risky I think she might be. She’s come here wanting to feel safe, and I’m here to help.
The views expressed here are entirely those of the author.
In the US, the National Suicide Prevention Lifeline is 1-800-273-8255
In the UK and Ireland, the Samaritans can be contacted on 116 123 or email firstname.lastname@example.org or email@example.com
In Australia, the crisis support service Lifeline is 13 11 14
Other international helplines can be found at www.befrienders.org