assessment of risk of harm to self or others

[sockit]

In practice - that is, in the practice of the various care providers typically encountered by someone with mental illness, i.e., GPs, nurses, psychiatrists, and social workers as well as psychologists - does the screen for harm amount to more than blunt questions like "are you thinking of hurting yourself/others"? What other information do practitioners use to make this judgement call in a real-world context?

(I would imagine a lot of patients might not be motivated to be honest if asked in that way, for a range of reasons.)

---

Members don't see this ad.

 

[PsyDr]

Like the movie ronin: "whenever there is any doubt, there is no doubt"

 

[LucidMind]

I don't mean to hi-jack the thread, but I have a related question. What does a screen for dangerousness look like?

 

[LucidMind]

In practice - that is, in the practice of the various care providers typically encountered by someone with mental illness, i.e., GPs, nurses, psychiatrists, and social workers as well as psychologists - does the screen for harm amount to more than blunt questions like "are you thinking of hurting yourself/others"? What other information do practitioners use to make this judgement call in a real-world context?

(I would imagine a lot of patients might not be motivated to be honest if asked in that way, for a range of reasons.)

Check out the suicide assessment proposed by Joiner, Walker, Rudd, & Jones (1999). Scientizing and routinizing the assessment of suicidality in outpatient practice. Professional Psychology: Research and Practice, 30, 1-7

 

[sockit]

psycscientist, thank you for your comprehensive and frank reply. It's good to know that some correlates that aren't related to self-report are at least sometimes taken into account.

PSYDR, I believe your doubting isn't to be doubted 🙂 but it seems like confidence in a given provider's clinical judgement might not be always be justified wrt people with different training, experience, and skills, in different settings.

LucidMind, thanks for the recommendation, I'll have a look.

Interested for sure in approaches that do not rely so much on explicit self-report. Epidemiological correlates can't be highly predictive on their own. I did find a recent paper in which researchers did a linguistic analysis of practioners' notes (Predicting the Risk of Suicide by Analyzing the Text of Clinical Notes), also not so strong, but interesting, I think. Would be interested too in similar analyses of transcripts of past therapeutic interviews/sessions with people who've completed suicide or homicide. But, seems that kind of research is preliminary and so far mostly equivocal. Hard to see how any applications would practically work in existing health care systems, given (possibly) fragmented information flows, variable training, etc. And some patients only see psychiatrists or GPs twice a year and get no therapy at all.

(I am thinking the Google of a few years ago might have been able to make a fair prediction based on search terms, for all it would help in practice, given that the data would have to be preemptively attached to an identifiable individual. People hate Google enough already, with good reason.)

 

[KillerDiller]

Check out the CASE approach developed by Dr. Shawn Shea (and others). I've found that approach to be extremely well thought out and supported method for eliciting self-reported suicidal ideation from clients. As far as considering other factors that aren't necessarily directly reported, I look at substance use, social support, isolation--basically the factors involved in the IS PATH WARM acronym. That being said, if someone is directly denying that they have homicidal or suicidal ideation (and is not experiencing psychotic symptoms), they are unlikely to be admitted to a hospital.

 

[WisNeuro]

Yeah, this question is one that is not purely clinical in nature. We have to take into consideration legal and ethical principles as psych says. There is a fine balancing point of sensitivity and specificity. Also, hospitalization even when you know risk is high can be tricky. If someone does not want to be hospitalized, and they know what to say (e.g., deny active plan, deny means, etc) then they can quite easily avoid an involuntary hold. Some states have a three-party override to that, but at most you'll get 3 days on someone who doesn't want to be there and has the insight to do what it takes to get out.

*These are extreme examples, I've worked with a fair share of individuals with Borderline PD. Luckily this is a low base-rate behavior, but still serious enough to make sure you do the legwork when dealing with suicidality.

 

[smalltownpsych]

A big factor in any risk assessment that is often overlooked is the clinician's ability to gain rapport. My supervisor and I used to have a friendly competition to see who could get the "truth" out of clients at an inpatient setting where I had a practicum. It was a good opportunity for that since we both had sessions with the individuals and we could compare notes. For me, it all boils down to building trust and when patients trust me, then they are more likely to tell me the truth regardless of what the question is.

Oh and I also agree with the assessment of risk factors vs. protective factors in the analysis.

 

[sockit]

Thank you all for your insights - I appreciate the discussion. It's a difficult problem, and there are clearly many issues to consider (that I haven't).

I was thinking along the lines of not necessarily hospitalization immediately but some non-human algorithm that might raise a flag for a harried GP who might miss signs, or for a psychiatrist who might interpret e.g. agitation as a simple side effect of a new drug, to tell him/her, "hey, take a second look here". But from a practical point of view, that would require integrated information perhaps across institutions, so no way to that; ethically it would certainly compromise privacy.

Re rapport - from what I have been reading and hearing, sometimes, in some settings, the key questions are asked after something like "are you allergic to any medications" - no rapport yet, no chance of it. But then again, if things are happening that way, not sure the practitioner asking the questions would be equipped (because of lack of training, or lack of time or resources) to take things in a helpful direction after that.

And of course the question of what actually can be done if suicidal/homicidal inclination is detected, to what degree - e.g. what a temporary hold would accomplish - is something else entirely.

*also there would be a danger to the patient who'd be followed by a fat file and a bunch of labels or indicators that might stick to them, which have not even been looked at by human eyes, but might in future.

 

[AcronymAllergy]

Just to speak to a general, anecdotal experience--I've actually found that for many things (including suicidal ideation), coming right out and asking about it (in an empathic, non-threatening, non-accusatory tone) tends to be appreciated by the patient as opposed to speaking in generalities, beating around the bush, etc. Although as smalltownpsych points out, rapport is a crucial contributing component.

 

[WisNeuro]

Just to speak to a general, anecdotal experience--I've actually found that for many things (including suicidal ideation), coming right out and asking about it (in an empathic, non-threatening, non-accusatory tone) tends to be appreciated by the patient as opposed to speaking in generalities, beating around the bush, etc. Although as smalltownpsych points out, rapport is a crucial contributing component.

From my recollection, the research supports this as well. Coming right out and asking is the best way. Also, as per the BDI, having an active plan, and hopelessness are the main predictors of suicide attempts. I'll try to track down the citation later.

 

[sockit]

I have not done an extensive amount of research (clearly), and I think it is obvious I have no personal experience, but the literature reviewed in that paper I linked to found different results:

"Clinicians generally ask patients whether they are “suicidal” and base their risk assessments primarily on the response. The concept of suicidality includes both thoughts about suicide and intentions to act on those thoughts [3]. While suicidality is a prominent risk factor for suicide attempts and completions, only approximately 30% of patients attempting suicide disclose their suicidal ideation [4], [5], [6], and the vast majority of individuals who express suicidal ideation never go on to attempt suicide [7], [8], [9]. Given this poor predictive value, clinicians might consider a more comprehensive approach by evaluating additional demographic risk factors for suicide."

these are the cited references

1. 4. Denneson LM, Basham C, Dickinson KC, Crutchfield MC, Millet L, et al. (2010) Suicide risk assessment and content of VA health care contacts before suicide completion by veterans in Oregon. Psychiatr Serv 61: 1192–1197. doi: 10.1176/appi.ps.61.12.1192
   • View Article
   • PubMed/NCBI
   • Google Scholar
2. 5. Kaplan MS, McFarland BH, Huguet N, Valenstein M (2012) Suicide risk and precipitating circumstances among young, middle-aged, and older male veterans. Am J Public Health 102 (Suppl 1): S131–137. doi: 10.2105/ajph.2011.300445
   • View Article
   • PubMed/NCBI
   • Google Scholar
3. 6. Kovacs M, Beck AT, Weissman A (1976) The communication of suicidal intent. A reexamination. Arch Gen Psychiatry 33: 198–201. doi: 10.1001/archpsyc.1976.01770020042006
   • View Article
   • PubMed/NCBI
   • Google Scholar
4. 7. Borges G, Angst J, Nock MK, Ruscio AM, Walters EE, et al. (2006) A risk index for 12-month suicide attempts in the National Comorbidity Survey Replication (NCS-R). Psychol Med 36: 1747–1757. doi: 10.1017/s0033291706008786
   • View Article
   • PubMed/NCBI
   • Google Scholar
5. 8. Crosby AE, Han B, Ortega LA, Parks SE, Gfroerer J, et al. (2011) Suicidal thoughts and behaviors among adults aged >/ = 18 years–United States, 2008-2009. MMWR Surveill Summ 60: 1–22.
   • View Article
   • PubMed/NCBI
   • Google Scholar
6. 9. Kessler RC, Berglund P, Borges G, Nock M, Wang PS (2005) Trends in suicide ideation, plans, gestures, and attempts in the United States, 1990–1992 to 2001–2003. JAMA 293: 2487–2495. doi: 10.1001/jama.293.20.2487

 

[AcronymAllergy]

A couple points (and admittedly I haven't looked at the study):

1) Context of the evaluation is going to be important
2) I've never asked a patient if they're "suicidal." Not once. By "explicit," I meant directly asking them if they're having thoughts about wanting to hurt themselves and/or are worried they might do so, if they're feeling hopeless, etc. I can see how folks would be turned off if I potentially tacitly labeled them as suicidal by the way I was asking my questions. This can be compounded by #1; if a patient is already feeling hurried through a primary care visit, for example, I'd imagine very few would have a positive reaction to then being asked if they're suicidal, or if it seems that their provider is simply checking off another item on a patient interview

Although yes, in general, we're pretty darn bad at predicting these things, which is helped in no small part because it's such a low base-rate behavior (thankfully) coupled with the idea that so much is based on self-report (and the timing of obtaining said self-report).

 

[sockit]

however it looks like the first two articles only looked at populations of veterans, who might have some special characteristics. and one of those references is from 1976.

 

[smalltownpsych]

I don't see how that equates to anything different from what you are being told about evaluating risk factors and protective factors along with asking the question in a direct manner. Also, asking the question in a direct manner helps build rapport, in my experience because most people beat around the bush and avoid which tends to increase the patient's perceptions that people don't care. The risk factors vs. protective factors help to guide me in my decision making process and also help craft the safety plan.

 

[smalltownpsych]

A couple points (and admittedly I haven't looked at the study):

1) Context of the evaluation is going to be important
2) I've never asked a patient if they're "suicidal." Not once. By "explicit," I meant directly asking them if they're having thoughts about wanting to hurt themselves and/or are worried they might do so, if they're feeling hopeless, etc. I can see how folks would be turned off if I potentially tacitly labeled them as suicidal by the way I was asking my questions. This can be compounded by #1; if a patient is already feeling hurried through a primary care visit, for example, I'd imagine very few would have a positive reaction to then being asked if they're suicidal, or if it seems that their provider is simply checking off another item on a patient interview

Although yes, in general, we're pretty darn bad at predicting these things, which is helped in no small part because it's such a low base-rate behavior (thankfully) coupled with the idea that so much is based on self-report (and the timing of obtaining said self-report).

At a hospital where I worked, the staff had a five question risk assessment that they would ask in a checking the box manner. I would meet with a patient right after they said no to all five checks, then the patient would tell me what was really going on. The science can guide us on the best approaches and the factors to evaluate, but it ultimately comes down to one human being in the room talking to another and that is where the craft and experience comes in to play.

 

[sockit]

Those are good points, and you're right, it doesn't. But I am not thinking about the psychologists who have the training and experience and sensitivity and judgement and time to think to weigh all these factors, develop rapport, etc; I am thinking about how the evidence can inform guidelines/algorithms to support GPs, social workers, etc.

 

[sockit]

I can see your point about the dangers of an algorithm. (And in that paper, for example, they used notes taken just from VA centres; there's no way to know that anything coming from that research would be useful wrt different institutions or different populations. Probably it would take a decade's worth of data to build a reasonably representative sample, by which time its characteristics might change... )

Are you kidding me, about GPs?

 

[WisNeuro]

Are you kidding me, about GPs?

I wish he were. I've been disappointed at best by most non-psychiatry MD's and how they approach suicidality and depressed pts.

 

[smalltownpsych]

Yes, but they're understandably uncomfortable about it. Once you ask the question, you have to do something. How is a GP going to do that without either 1. psychiatric training or 2. guaranteed back-up?

If you're interested, here is a recent paper published in response to resistance over a bill in Washington State that would mandate suicide risk training for primary care docs, among others:
http://onlinelibrary.wiley.com/doi/10.1111/sltb.12010/abstract

Also, interesting anecdote from the Henry Ford System: they were able to overcome resistance to the idea of asking about suicide in GPs when they were guaranteed a psyc consult within 24 hours (see p4):
http://actionallianceforsuicideprev...taskforces/ClinicalCareInterventionReport.pdf

Maybe they would be more comfortable asking the questions if we had better relationships with them. One area that I think that the APA is pushing in the right direction.

 

[sockit]

Very interesting links, thank you - will be reading for sure. I am definitely interested in systems-level issues.

Thanks to everyone who contributed to this thread. I've found it hugely informative.

 

[AcronymAllergy]

Having mental health integrated into primary care (such as with what the VA has been pushing for a few years now) is definitely an improvement, IMO. That way, if the primary care doc is worried, they can just have the person either seen for a walk-in there in the clinic, or immediately scheduled for a follow-up assessment at a later date. Takes the onus for more in-depth assessment off of them and eases the referral process/transition.

 

[vesper9]

I am thinking about how the evidence can inform guidelines/algorithms to support GPs, social workers, etc.

2 different issues here:
1) an algorithm to calculate risk
2) how to handle that risk

There are already algorithms that exist. Google P4 screener for an example. It asks several questions about the components of suicide risk (history, plan, intent, preventative factor) and triages the client to one of 3 levels of risk (minimal, lower, higher).

This is useful for things like research surveys or symptom measures that ask about self-harm conducted by non-clinicians. That way the initial triage of risk level is not subjective and dependent on the experience of the interviewer.

BUT if a persons's risk screen is in the higher level, you will still need them to be assessed with a trained mental health clinician that can more finely assess the risk through interview and determine an appropriate intervention (like everyone mentioned above).

Most suicidality interventions probably don't involve hospitalization, but safety planning around eliminating means and/or increasing protective factors. Trained and experienced mental health providers are needed for that.

 

[sockit]

Of course - I'm not arguing for the elimination of psychologists, and I don't imagine I'll invent anything myself 🙂

The P4 screener looks similar to the general approach others have mentioned, though, is that right? What I find interesting about the approach in the study above is (other than the fact that I lucked into it 😉 so am kind of attached to it now), it's more like data-driven factor analysis, which might pick up something unexpected (or might not, I don't know). E.g. the word graph here, showing frequencies of phrases that appeared in notes about people who actually completed suicide, shows some things that may be obvious to a clinician, but might offer neat insights. Lots of people who completed suicides at the VA centres studied were "agitated" and afraid (or were themselves actively "frightening", don't know if that is a direct quote) and psychotic ("delusional"); many appear to have had pain ("analgesia", "demerol"); at least some had comorbid health issues ("vtach", "swabs", "secretions", "urologic") and interpersonal issues were prominent. All of these correlations might have already been seen in existing epidemiological studies, but it's interesting nonetheless, I think.

 

[AcronymAllergy]

It's an interesting approach, and I honestly haven't yet read the study (so they might've addressed this), but one thing to keep in mind is that especially at the VA, notes can vary SUBSTANTIALLY from provider to provider (even with the same patient). Thus, unless the providers were asked to use some sort of template, to include the same sorts of information, and to use the same terminology, it seems to me initially like a lot may be simply based on coincidence and/or inadvertently on clinical judgment (e.g., "I'm going to include this exact phrase the patient said and/or word it this way because something just felt off and I'm a bit worried about this guy").

Don't get me wrong, it's an interesting and novel approach, but I just wonder if it'd end up being useful above and beyond current methods (particularly considering they were working from a known-groups retrospective design). What could perhaps be more interesting would be voice and lexical analysis of actual session recordings, although again, there'd need to be a LOT of standardization that would take place across the therapists.

 

[DD214_DOC]

I use a crystal ball or some other form of divination. All of these intricate assessments are cute, but really only exist so that we can document that we did something we think actually matters for liability purposes. In reality, none of it matters and we cannot predict who may or may not do something. A person can just as easily say, "Nope, I'm not suicidal and never have been", then go home and kill themselves.