- Suicide risk assessment is a necessity in clinical practice, but "screeners" and "checklists" tend to have low positive predictive value.
- Predicting suicide with 100% certainty is not the goal of risk assessment; prevention is.
- Better suicide prevention efforts require addressing the root causes of distress, hopelessness, and despair.
This is Part 2 of my interview and dialogue with Dr. Tyler Black, a child and adolescent psychiatrist, suicidologist, and clinical assistant professor at the University of British Columbia.
Joe Pierre: Let’s talk about suicide risk assessment. Continuing in the vein of our last conversation that even experts are often bad at predicting the future, a number of important studies have shown that suicide risk-assessment tools that use actuarial risk factors to estimate suicide risk or to place people into low, intermediate, and high risk categories aren’t actually very good at predicting completed suicide, self-harm, or suicidal behavior.1-5 It’s often said that we can’t predict suicide and that it’s more like weather forecasting. Part of the reason seems to be that completed suicide is actually relatively uncommon and that many risk factors are common and unmodifiable (e.g. like being elderly, white, and male) or nonspecific such that they have low positive predictive value and potentially large numbers of false positives and false negatives. But even something as specific as suicidal ideation isn’t very predictive of suicidal behavior.6
As someone who has created their own suicide risk assessment inventory, what’s your take on the utility of such instruments as screening tools? How effective are they really and if they’re not that useful, why—or how—should we use them? Given how many different tools are out there for this purpose, which ones do you think are better or worse, and what makes them so?
Tyler Black: This is such a great question and framing. All (and I mean all) of the suicide risk prediction tools/scores out there are completely worthless as predictors. The problem is, in medicine, we are trained to separate "high risk" from "low risk." We do it with the Ottawa Ankle Rules, chest pain presentations, and pretty much every other health issue. In medicine, we are actually quite poor at our approach to things we can't predict. We act like we can, we go by algorithm, and we create policies of care, without any evidence behind it. This creates harm. About 99% of people screening positive for a "suicide screen" will not die by suicide, and yet many times their rights are taken away and treatment is forced on them (despite this same treatment not showing evidence of benefit for suicide outcomes). Worse, 50% of people who will die by suicide every year would screen as "low risk" on the common scales. With such a high false positive and negative rate, we have major issues in prediction.
The reason I created the Assessment of Suicide and Risk Inventory (ASARI) was to create a documentation tool to reflect the knowledge a clinician has. It is not intended to predict outcome or even drive treatment. Unfortunately, many administrators will still have "high risk" vs "low risk" patient policies. Heck, even entire programs will reject or accept patients based upon these fabricated risk assignments.
The gold standard for suicide risk assessment is "get to know your patient." Don't checklist them. Don't try to fit them into a box. Understand them. Find out what makes their life more stressful and what improves their life. What's missing? What's needed? What's good and needs to be supported? I don't particularly care if you use ASARI, the Columbia-Suicide Severity Rating Scale (C-SSRS), the Ask Suicide-Screening Questions (ASQ), the LMNOP (I made this up) or any other scale. None of them are predictors and none of them can replace the work of getting to know your patient.
A proper suicide risk assessment includes:
- A detailed analysis of the risk and protective factors present in the patient
- An interview in which the clinician and the patient can have a dynamic interaction so the clinician can better understand the above
- A synthesis and common understanding ("formulation") of the above
- An approach to reducing risk by reducing established/possible risk factors and bolstering established/possible protective factors
That's it. That's the magic. Everything else is just bureaucracy and hoops. Good clinicians should fight the "checklisting" of this.
JP: On the one hand, research suggests that suicide risk assessment scales aren’t any better than the global judgment of experienced clinicians. But other research has suggested that extracting data from electronic health records and using digital interventions for suicide preventing might be promising directions for getting better at predicting and preventing suicide.7,8 As a guy who is optimistic about technology, how do you see “digital strategies” impacting our ability to reduce suicide in the future, particularly among young people for whom being online or on social media might worsen suicide risk?
TB: I think there can be good work with identifying clusters in this space (especially as we improve the speed of reporting), but in terms of predicting on the individual level, I don't think it's within our current reach. I hope to be proven wrong, but the devil is always in the "positive predictive value" details. The creators of these things love to show specificity and sensitivity. Can I see technology providing me an insight or pattern that maybe I didn't pick up on? Sure. But like everything else, it's going to have a lot of false positives.
JP: Expecting that risk assessment would predict suicide with a high degree of certainty would also mean that it’s clinically useless because the goal is really prevention, not prediction. But on another discouraging note, a recent meta-analysis concluded that no suicide prevention strategy is particularly effective in terms of reducing self-injurious behavior and that we haven’t gotten much better at preventing suicide compared to what we were doing 50 years ago.9 However, other reviews suggest a variety of interventions—ranging from psychoeducation to suicide safety planning to preventing access to firearms—that can reduce suicide-related outcomes.10,11 What’s your perspective on how to resolve these findings? What do you think really works and doesn’t work in terms of suicide prevention?
Some have suggested that we need to get better at understanding actual causes of suicide rather than risk factors. When I conceptualize suicide, I borrow from Edwin Shneidman’s portrayal of it as an alternative that people choose when their lives are intolerable. It has similarly been said that “when you don’t have any doors to open, death is a door.” To me, this means that prevention should mostly be focused on how to open doors so that people’s lives become tolerable or even good.
In the patient population that I work with, addressing homelessness and substance use disorders is a huge part of that. From both a general population perspective and as a child psychiatrist working with young people, what’s your take on where to invest most heavily in strategies that will actually reduce rates of completed suicide?
TB: Well, we have to agree on the goal. "Zero suicides" is impossible and probably not even really desirable. I personally support “dignity with dying” and medical assistance in dying and I am a full-time suicidologist who tries to help many, many suicidal people. So, as soon as we add that nuance, my position aligns with what you've said about cause. We have to address the root causes of distress, hopelessness, and despair if we are serious about reducing suicides. Hotlines and hospitals are like the basket at the bottom of the waterfall. Reducing child abuse will reduce suicide rates. Ending sexual assault will reduce suicide rates. Reducing barriers to disabilities will reduce suicide rates. These are huge-ticket items, and they are far more costly and complex than "put a phone on a bridge."
The evidence on means reduction is relatively robust, and I think that whenever possible, gun access, bridge access, and train access should be reduced. Of course, there are other ways to impulsively die but with only a few exceptions, removing impulse-possible hazards from the environment has reduced suicide rates.
I do think that emergency lines have a place—when someone is in crisis I'm so glad they have someone to call. But, by definition, one has to be at least somewhat ambivalent about dying to make the call in the first place. There is zero value to calling a crisis line if you are 100% intent on dying; it can only interfere with motivation. This aligns with what we know about death statistics—most who die did not call a hotline, and most die on the first attempt (rather than subsequent attempts).
If we are the basket at the bottom of the waterfall, we are missing a lot of tumbling people.
See Part 1 of my interview with Dr. Black: "Suicide in the Pandemic: A Prediction That Didn't Come True."
If you or someone you love is contemplating suicide, seek help immediately. For help 24/7 contact the National Suicide Prevention Lifeline, 1-800-273-TALK, or the Crisis Text Line by texting TALK to 741741. To find a therapist near you, see the Psychology Today Therapy Directory.
1. Chan MKY, Bhatti H, Meader N, et al. Predicting suicide following self-harm: systematic review of risk factors and risk scales. British Journal of Psychiatry 2016; 209:277-283.
2. Quinlivan L, Cooper J, Davies L, et al. Which are the most useful scales for predicting repeat self-harm? A systematic review evaluating risk scales using measures of diagnostic accuracy. BMJ Open 2016; 6:e009297.
3. Carter G, Milner A, McGill K, et al. Predicting suicide behaviours using clinical instruments: systematic review and meta-analysis of positive predictive values for risk scales. British Journal of Psychiatry 2017; 210:387-395.
4. Quinlivan L, Cooper J, Meehan D, et al. Predictive accuracy of risk scales following self-harm: multicentre, prospective cohort study. British Journal of Psychiatry 2017; 210:429-436.
5. McHugh CM, Corderoy A, Ryan CJ, et al. Association between suicidal ideation and suicide: meta-analyses of odds ratios, sensitivity, specificity and positive predictive value. BJPsych Open 2019; 5, e18.
6. Franklin JC, Ribeiro JD, Fox KR, et al. Risk factors of suicidal thoughts and behaviors: a meta-analysis of 50 years of research. Psychological Bulletin 2017; 143:187-232.
7. Simon GE, Johnson E, Lawrence JM, et al. Predicting suicide attempts and suicide deaths following outpatient visits using electronic health records. American Journal of Psychiatry 2018; 175:951-960.
8. Torok M, Han J, Baker S, et al. Suicide prevention using self-guided digital interventions: a systematic review and meta-analysis of randomized controlled trials. Lancet Digital Health 2020; 2:e25-36.
9. Fox KR, Huang X, Guzman E, et al. Interventions for suicide and self-injury: a meta-analysis of randomized controlled trials across nearly 50 years of research. Psychological Bulletin 2020; 146:1117-1145.
10. Hofstra E, van Nieuwenhuizen C, Bakker M, et al. Effectiveness of suicide prevention interventions: a systematic review and meta-analysis. General Hospital Psychiatry 2020; 63:127-140.
11. Mann JJ, Michel CA, Auerbach RP. Improving suicide prevention through evidence-based strategies: a systematic review. American Journal of Psychiatry 2021 (in press).