Top 5 mistakes in suicide risk assessment
Brainless errors and their thoughtful fixes
Mistakes are OK, and in clinical settings, they happen regularly. You assume something, get it wrong, and learn that assumptions are dangerous. You skip a step and an untoward event happens. You misspeak and end up offending someone. In most situations, we quickly identify mistakes because of swift or real-time feedback. However, suicide risk assessment is a really interesting problem: >99% of the people we see for suicidal thinking will not die of suicide, and >99% of the people we hear about who die of suicide we will not see clinically. This puts the suicide risk assessor in a bit of a problem when it comes to self-editing, so it’s a judgment-free zone if you read this and you think, “huh, I do that!”
In the examples below, we’ll follow the documentation on suicide risk about a fictional 65-year-old man named Joseph, and how a brainless and thoughtful suicide risk assessment can lead to wildly different approaches.
“Success does not consist in never making mistakes but in never making the same one a second time.”
― George Bernard Shaw
Mistake #1 - Not thinking deeply about the “suicidality” screening question.
Many clinicians quickly learn from a colleague, mentor, or other’s notes to always document the answer to the presence suicidality, and the documentation is something like this: “<name> denies suicidality.” This is not an unimportant question, but in a very meaningful way, it is completely unimportant. Let me explain.
We care very much about our patient’s point of view (as a physician, I use the term patient, and despise the term ‘client’ or ‘service user,’ but this is not my fight today so please forgive my use of this word). When it comes to suicide risk assessment, however, we have a very important paradox.
If a patient says they are not suicidal, either is true:
They are not thinking about suicide and telling you OR
They are thinking about suicide and not telling you
There is no discernable way to identify the two, and there is another wrinkle to add: if a patient 100%, without question or any ambiguity, wants to die by suicide, a clinician, assessor, physician, counsellor, friend, family member is a thwarter to that suicide.
If a patient says they are suicidal, either is true:
They are thinking about suicide and telling you OR
They are not thinking about suicide and not telling you
There is no discernable way to identify the two, and there is another wrinkle to add: if a patient 100%, without question or any ambiguity, does not want to die but wants to ensure that a clinician, assessor, physician, counsellor, friend, or family member takes their concerns and life seriously, ‘suicide’ is a word that recruits immediate and prompt concern.
In 2023’s health care and mental health care climate, in fact, being suicidal is requisite for access to many emergency services.
In other words, for both the “I’m not suicidal” and “I’m suicidal” answer, there are incentives that could hide the true answer, and there is no way to tell what someone’s true intentions or thoughts are in a definitive way.
The brainless documentation of “denies suicidality” or “endorses suicidality” hides this issue from both the writer AND the reader.
The fix is very simple: consider and document their answer, and then immediately pair it with what you know and think about that answer. Did it seem like it fit with the rest of your assessment? Was it something that made you feel that perhaps you were missing something? Did your gut tell you that something is amiss? Document that!
Brainless example: Joseph denies suicidality.
Thoughtful example: Joseph denies suicidality which is reinforced by his orientation to his future plans to graduate, his expressed desire to seek care for his stressors, and on the clinical interviewer my examination revealed an interactive and thoughtful discussion about risk including steps forward.
Mistake #2 - Including demographics (age, racial, sex, orientation), past attempt, or psychiatric history based risk factors in your assessment of acute suicide risk
We all know the stats. In most countries, males outnumber females for suicide. White Americans die by suicide rate at higher rates than Black Americans do (But less than Indigenous Americans). Elderly men die of suicide at higher and higher rates as their age pushes past 60. The most accurate predictor of future suicide is previous suicide attempts. This is important public health information and represents the bulk of our “well established knowledge” on “risk factors” (which is inherently problematic and denies systemic and social pressures on these groups).
So, reader who is a clinician, here’s a question I bet you didn’t expect: so bleeping what?
If I’m assessing a 13-year-old girl (one of the lowest rates in Canada) and she has no prior psychiatric history or self harm history (knocking out two of the strongest predictors), who is seeing me because of suicidal ideation, I do not benefit whatsoever by considering her sex or age when it comes to her clinical risk of suicide. I am not comforted by the lack of prior history because from this point forward, she now has “suicidal ideation” in her history!
Of course, she may experience issues due to her age (child abuse, online exploitation, social media concerns, educational concerns, bullying) or sex (misogyny creating realistic barriers to her endeavours, sexual pressures or violence, health care minimization) that may be of paramount concern to me, but I don’t think to myself “yeah, sure she’s saying she’s suicidal - but she’s 13 and a girl so she’s at low risk.”
The fix is obvious: put everything you know about demographics and historical risk factors, and put them in the “chronic risk” bucket. They elevate the baseline, but you have no idea about the relative risk after that. My ASARI does a great job of separating chronic baseline issues (that are unmodifiable and/or historical) from acute issues (that can change and/or are new). As soon as you start seeing an individual, you have to throw your demographics out the window, and respond to what the person is facing and what is increasing and decreasing their relative risk.
Brainless example: Joseph is 65 years old and male, with a previous suicide attempt and a prior diagnosis of depression, so I am concerned about his suicidality.
Thoughtful example: Joseph, a 65-year-old male, is presenting with anxiety and has a prior attempt of suicide in his remote history. Though these demographics are over-represented in suicide deaths, Joseph himself seems to have a rewarding life that he enjoys, and his anxiety seems to be his primary concern for this assessment. He tells me that psychotherapy in the past has helped him with his struggles and he has not had suicidal thinking since.
Mistake #3 - Caring about stratification
In many jurisdictions and hospitals, it is a requirement to document “high” or “low” risk of suicide for various procedures and policies to come into effect. This evidence-free implementation is likely more harmful and clinically useless.
The vast majority of people who die of suicide die on their first attempt, so “prior attempt” has a very underwhelming 10-year positive predictive value (4%). The vast majority of people who are assessed to be “high risk” for suicide will not die of suicide (95+% over 5 years ). Nearly 10-50% of all suicide deaths occur in people with “low risk” (machine algorithms are improving this, but it’s always the tension between reducing false positives by increasing specificity and reducing false negatives by increasing sensitivity).
We are not yet in a world where there is a clinically meaningful “high risk” group, yet many systems and clinicians act like we are. We use “high risk” as a measure to prioritize clinical care (people low risk get to wait longer!), or reasons to take away people’s rights through involuntary commitment, or other things.
The fix is very simple - the gold standard for suicide risk assessment is a clinician’s judgment through the gathering of information and interview of the patient. I don’t care if my patient scores low on a scale or someone else says they’re low risk; if I identify something that is a risk factor to increase their chance of suicide, or a missing protective factor, I go for it. On units, wards, and health care teams, we need to discourage stratification as a means for accessing services. This creates poor incentives in people seeking care and neglects the well-held truth that important early issues might be the most important and clinically meaningful way to change trajectory.
Brainless example: Joseph scores high on SADPERSONS [ugh, don’t get me started. NEVER USE SADPERSONS] so is deemed a high suicide risk.
Thoughtful example: Many of Joseph’s demographics and histories score on the SADPERSONS [ugh, remember what I said? don’t use it!] scale, however the clinical interview, reason for presentation, and my mental status examination reveal someone with only their baseline levels of suicide risk. I am most hopeful that our treatment plan below is the proper approach to reducing their relative suicide risk, as anxiety seems to be their pressing problem.
Mistake #4 - Confusing “what happened” for “why”
(this excludes our Joseph example, but it’s so common I have to mention it)
A new psychologist is asked to assess someone after a suicide attempt. They document:
When asked why they took pills, they told me they had a fight with their girlfriend over finances.
Nope. That’s just a longer form of “what happened.” It’s a relevant stressor and something to think about. Why is deeper:
When asked why they took pills, they told me they had a fight with their girlfriend over finances, and believed that their girlfriend would leave them. They love their girlfriend and catastrophized that they would split up over this, though they now feel sheepish about this because their girlfriend is with them today and the financial stress was resolved.
In the latter example, we get an understanding of why. They feared being alone, and the prospect of being alone was so painful to them that they’d rather die. They catastrophized the argument, showing that in that high-pressure situation, they had a hard time thinking clearly about it and were inaccurate in arriving to the correct conclusion. They feel sheepish about their attempt because they recognize that it was a really poor decision. I am describing my modern version of Durkheim’s bizarrely worded anomic suicidal motivation: the chaotic approach to stress that results in poor planning, over-reaction, and impulsive decisions. The best thing about understanding why is it previews a way to help. Someone in this situation would benefit very well from understanding their catastrophic reaction, how to communicate and receive communication in more productive ways, and safety planning for the next time they feel overwhelmed.
Mistake #5 - Making complex assessments seem simple
I cannot tell you the number of physicians, clinicians, counsellors, and mental health staff who feel that the phrase “denies suicidality” is a sufficient medicolegal documentation. Due to the explanation I provided above, “denies suicidality” is medicolegally meaningless. The most incompetent lawyer in the world could ask you on the stand “So, has it ever occurred that someone who died of suicide denied suicide? How often has that happened?” and then you would immediately realize that you have absolutely zero shield by your meaningless documentation.
If you have ever wondered where medicolegal risk lies, it’s in the following question: “Did you meaningfully assess and consider this person’s suicide risk?”
If you write “denies suicidality” OR “endorses suicidality” in your document and only that, you have very little evidence that you meaningfully considered the risk of suicide.
Brainless example:
Joseph denies suicidality.
Thoughtful example - Here’s what we’ve written so far based upon my “fixes”
Joseph is a 65-year-old male, and is presenting with anxiety and has a prior attempt of suicide in his remote history. Though these demographics are over-represented in suicide deaths, Joseph himself seems to have a rewarding life that he enjoys, and his anxiety seems to be his primary concern for this assessment. He tells me that psychotherapy in the past has helped him with his struggles and he has not had suicidal thinking since. Joseph denies current suicidality which is reinforced by his orientation to his future plans to graduate, his expressed desire to seek care for his stressors, and on the clinical interviewer my examination revealed an interactive and thoughtful discussion about risk including steps forward.
Many of Joseph’s demographics and histories score on our department’s clinical scale, however the clinical interview, reason for presentation, and my mental status examination reveal someone with only their baseline levels of suicide risk. I am most hopeful that our treatment plan below is the proper approach to reducing their relative suicide risk, as anxiety seems to be their pressing problem.
Which do you think is more medicolegally protected?
Conclusion: Brainless suicide risk assessment benefits nobody, and unfortunately, is often encouraged by systems and their checklists, charts, and scoring systems. Be thoughtful in your suicide risk assessment and you will sleep better at night, knowing that you’re truly doing your best to help and understand people.
Over the course of my writing here, I hope to share more clinically useful pearls. Please let me know in the comments or over on the ol’ twitterbird and i’d be happy to answer any questions.
Ok, but serious question, why are you the only person on the planet who writes clinically meaningful information about suicide risk? Every time! Thanks Tyler.